| mordred | however, I'm going to call bs on anyone being confused by the "put requirements in requirements.txt" part | 00:00 |
|---|---|---|
| lifeless | And as dstufft says, having a clear and straightward description of how to translate can be done even before that nirvana is reached | 00:00 |
| mordred | it's a clear and simple delta | 00:00 |
| mordred | "put what you put in setup.py into setup.cfg except for requirements which you put in requirements.txt" | 00:00 |
| mordred | done | 00:00 |
| lifeless | except extras | 00:00 |
| lifeless | and setup requires | 00:01 |
| mordred | nobody | 00:01 |
| mordred | NOBODY | 00:01 |
| mordred | understand those | 00:01 |
| dstufft | I do! | 00:01 |
| mordred | except liek 4 of you | 00:01 |
| dstufft | :D | 00:01 |
| mordred | dstufft: you don't count | 00:01 |
| lifeless | actually | 00:01 |
| lifeless | there's a fair whack of packages using them | 00:01 |
| lifeless | I noticed that when I was writing my map-reduce over pyPI for setup-requires usage | 00:01 |
| mordred | there's also a fair whack fo packages compiling C | 00:01 |
| mordred | doesnt' mean I want to join their crazy | 00:01 |
| mordred | so - let's constrain the conversation real quick | 00:02 |
| mordred | because I tink I am understanding your POV better | 00:02 |
| mordred | the semantics of an existing requirements.txt in the absence of install_requires in setup.cfg CANNOT change | 00:02 |
| mordred | that's non-negotiable | 00:02 |
| mordred | because of backwards compat | 00:02 |
| lifeless | agreed - I specified for that in my email | 00:02 |
| mordred | yah - just being clear here in channel | 00:03 |
| lifeless | there's about a dozen existing files covered by that statement | 00:03 |
| mordred | moving forward - putting requirements into setup.cfg is totally doable | 00:03 |
| mordred | and I have no personal problem with it - nor do I think it will be particularly confusing to people | 00:03 |
| lifeless | (all the-PYN ones,and the alternative paths) | 00:03 |
| mordred | lifeless: tangent - we should remove *-requires from teh docs but not the code) | 00:04 |
| mordred | we support them for history - but really don't want to add new ones | 00:04 |
| mordred | or something | 00:04 |
| mordred | mark the in the docs as historical abberations | 00:05 |
| mordred | from the same people who brought us "rpms" vs. "apts" :) | 00:05 |
| *** achanda has quit IRC | 00:05 | |
| lifeless | yeah | 00:05 |
| lifeless | y | 00:05 |
| mordred | lifeless: do you feel strongly that you want to advocate for a behavior other than unsupported in the presence of both requirements.txt and install_requires ? | 00:06 |
| lifeless | hell no | 00:06 |
| mordred | great | 00:06 |
| *** jamielennox|away is now known as jamielennox | 00:06 | |
| *** davideagnello has joined #openstack-infra | 00:06 | |
| mordred | then I think we have no real problems | 00:06 |
| lifeless | I want both to mean 'use setup.cg and ignore requirements.txt.' | 00:06 |
| lifeless | but if other people feel strongly I'd be ok with | 00:06 |
| lifeless | 'error' | 00:07 |
| mordred | I think I'd like to just not let that condition exist in openstack | 00:07 |
| lifeless | so lets look at that | 00:07 |
| mordred | and if the impl is "setup.cfg overrides requirements.txt" | 00:07 |
| mordred | it won't really matter, since we'll have no repos in that state | 00:07 |
| lifeless | if we have something we're deploying, and we want a requirements.txt for the deploy - to pass to pip specifically. | 00:07 |
| lifeless | what would we do ? | 00:07 |
| lifeless | e.g. nodepool. | 00:08 |
| mordred | I do not understand why they would be different | 00:08 |
| mordred | i.e. I reject the premise that that's useful | 00:08 |
| mordred | nodepool will have its deps list in setup.cfg and pip install nodepool will be how you install nodepool | 00:08 |
| mordred | if that doesn't work, then something is wrong with the model | 00:09 |
| lifeless | say there's a new release of a dep that breaks nodepool.. Would you commit an exclusion to nodepool, or would you override at install time | 00:09 |
| tchaypo | in the discussion about stable branches, one case that was mentioned was if one of nodepool’s dependencies has had a bug introduced in a minor release | 00:09 |
| mordred | yes | 00:09 |
| mordred | I would commit an exclusino | 00:09 |
| lifeless | thats where AIUI requirements.txt is intended to exist. | 00:09 |
| mordred | beacuse otherwise nodepool is broken | 00:09 |
| mordred | it is quite literally a bug in a nodepool install if that situation exists | 00:09 |
| tchaypo | we might catch it later in CI, but someone doing “pip install nodepool” has no way of knowing what versions we tested in CI | 00:09 |
| tchaypo | unless we record it somewhere (I’m not convinced requirements.txt is the right place to record that though) | 00:10 |
| mordred | tchaypo: that's why I've been arguing for == pins for everytihng - but I keep losing | 00:10 |
| mordred | tchaypo: because python breaks things on minors more often than is sane | 00:10 |
| mordred | and what I _can_ assert is that I have tested versions X, Y and Z | 00:10 |
| tchaypo | mordred: great, I know who to point the finger of blame at, which is nice - but what I really want is a way to get a running system *now* with a tested-and-working version | 00:10 |
| lifeless | so thats the thing - the split dstufft argues for is that a requirements.txt can have such pins | 00:10 |
| mordred | tchaypo: right | 00:10 |
| mordred | lifeless: but where does the requirements.txt come from? | 00:11 |
| mordred | how does it solve the problem any better? | 00:11 |
| dstufft | hopefully we can change that by pushing people to use ~= (and maybe a future ^=) and convince people to maintain some semblence of compataiblity | 00:11 |
| tchaypo | see my email for a spec.. | 00:11 |
| fungi | it's also worth remembering how we got to pbr too (tools/pip-requires files getting preinstalled into virtualenvs). pbr, while perhaps suboptimal, is an evolution of that | 00:11 |
| *** amitgandhinz has quit IRC | 00:11 | |
| mordred | if something breaks nodepool, we'd need to update the requirements.txt | 00:11 |
| mordred | so that users would know about it | 00:11 |
| mordred | why not just update setup.cfg? | 00:11 |
| mordred | like, why keep a separate list? | 00:11 |
| *** davideagnello has quit IRC | 00:11 | |
| tchaypo | mordred: for people doing downstream distributions, who might need to consider dependencies for things other than openstack | 00:12 |
| mordred | tchaypo: they're already covered | 00:12 |
| tchaypo | how? | 00:12 |
| mordred | tchaypo: pbr doesn't handled depends at all if you're doing debian packaging | 00:12 |
| mordred | tchaypo: because it's understood that the distro will | 00:12 |
| jamielennox | this might be what we are discussing now - but is there a fix for the ceilometerclient cap problem in juno branch? | 00:12 |
| mordred | jamielennox: it's not at all | 00:12 |
| mordred | and i have no idea | 00:12 |
| *** tiswanso has joined #openstack-infra | 00:13 | |
| tchaypo | jamielennox: I’ve been thinking about that over the weekend, and I think the solution there is to make the stable jobs non-voting for now | 00:13 |
| mordred | tchaypo: so - yeah - the downstream distros thing in your email is a red herring | 00:13 |
| tchaypo | as I don’t think there’s any way we can ever get consistency with the current jobs | 00:13 |
| mordred | tchaypo: pbr should not be doing anything with versions, tight or not, for downstream distributors | 00:14 |
| mordred | tchaypo: http://docs.openstack.org/developer/pbr/packagers.html | 00:14 |
| lifeless | mordred: requirements.txt is a better thing to edit that install_requires when you're installing a local version of a dep. | 00:14 |
| mordred | lifeless: let me reask the question | 00:14 |
| mordred | lifeless: why would we, as nodepool upstream, put a requirements.txt file into our repo? | 00:15 |
| jamielennox | grand scheme there are venvs and such that we could do to make it work, but it doesn't solve the issue of packaging (which i'm generally staying out of) | 00:15 |
| fungi | jamielennox: tchaypo: a bunch of lib/client reqs were capped in global-requirements.txt using <=(latest tag) which doesn't work when trying to test proposed changes on top of those stable branches since the'll get (latest tag).postN | 00:15 |
| lifeless | mordred: nodepool is in the special case of | 00:15 |
| lifeless | mordred: 'we write and CD it ourselves' | 00:15 |
| lifeless | mordred: its why I picked it | 00:15 |
| mordred | lifeless: I'm really trying to understand a concrete example of when a requirements.txt would want to go into a git repo | 00:15 |
| fungi | jamielennox: tchaypo: simple fix is to change those caps to strictly <(next semver release).0 | 00:15 |
| lifeless | mordred: so maybe it would go into system-config | 00:15 |
| mordred | lifeless: that's what I'd do for something I was deploying I didn't write | 00:16 |
| dstufft | generally upstream doesn't have a requirements.txt unless the expected way of installing something is "git clone and pip install -r requirements.txt" | 00:16 |
| mordred | and I'd expect other fokls to do similar | 00:16 |
| jamielennox | fungi: the one i'm getting now is pkg_resources.ContextualVersionConflict: (python-keystoneclient 1.3.0.post18 (/opt/stack/new/python-keystoneclient), Requirement.parse('python-keystoneclient<=1.1.0,>=0.10.0'), set(['ceilometer'])) | 00:16 |
| mordred | dstufft: if that is the expected way to install it, I'm not likely going to be installing it | 00:16 |
| lifeless | mordred: so if the fix for a nodepool issue was to install a local patched dep; how would you do that today? | 00:16 |
| fungi | jamielennox: yep, that's it | 00:17 |
| * SpamapS tries desperately to catch up to the conversation. | 00:17 | |
| mordred | lifeless: system-config | 00:17 |
| mordred | lifeless: although, I'd likely look in to dropping that depend | 00:17 |
| jamielennox | fungi: i had thought it was a ceilometerclient thing, but that might have been another i saw - i know bknudson was looking at something with keystoneclient caps | 00:17 |
| lifeless | mordred: so if we're willing to say that thats how we keep doing it, then we can say we don't ever need a pip-style requirements.txt in nodepool | 00:17 |
| fungi | jamielennox: oh, for that one i think cielometerclient may also have a src-juno job being tested on master. it should stop running that | 00:17 |
| lifeless | mordred: and then we're resolved on this subthread. | 00:18 |
| tchaypo | fungi: so that one should be python-keystoneclient<1.2.0 ? | 00:18 |
| mordred | lifeless: cool | 00:18 |
| fungi | jamielennox: what's the job name and the branch of the failing change? | 00:18 |
| *** yamada-h has joined #openstack-infra | 00:18 | |
| dstufft | gotta run to the store real quick | 00:18 |
| mordred | lifeless: I believe there is almost no work needed on pbr to support this | 00:18 |
| bknudson | jamielennox: I was looking at that in stable/juno... haven't had a chance to look a master lately. | 00:18 |
| jamielennox | fungi: that one is https://review.openstack.org/#/c/171462/ gate-tempest-dsvm-neutron-src-python-keystoneclient-juno | 00:18 |
| mordred | lifeless: it'll mostly be work to update.py in o/r | 00:18 |
| bknudson | jamielennox: that job should be gone now. | 00:18 |
| tchaypo | fungi: there’a thread with subject "novaclient 'stable-compat-jobs-{name}’ broken” with other examples | 00:18 |
| lifeless | mordred: its a shallow patch yes | 00:19 |
| mordred | lifeless: and documentation | 00:19 |
| tchaypo | gate-tempest-dsvm-neutron-src-python-novaclient-icehouse | 00:19 |
| tchaypo | gate-tempest-dsvm-neutron-src-python-novaclient-juno | 00:19 |
| jamielennox | bknudson: ah - ok, i rechecked it yesterday and just got around to looking now | 00:19 |
| mordred | lifeless: I believe we already support requires-dist - yeah? | 00:19 |
| lifeless | mordred: I started teh conversation so that if it turned out to be thorny we could make time in vancouver | 00:19 |
| mordred | nod | 00:19 |
| fungi | jamielennox: so, .*-src-python-keystoneclient-juno is the "juno backwards-compatibility integration testing for python-keystoneclient" | 00:19 |
| *** baoli has joined #openstack-infra | 00:19 | |
| mordred | lifeless: have brainspace for more pbr chatting while you're here? | 00:19 |
| lifeless | mordred: except everyone hates that key, so there's other work to support that :) | 00:19 |
| lifeless | mordred: I can sure | 00:19 |
| mordred | lifeless: ++ | 00:19 |
| mordred | lifeless: it's the one thing we can ALL agree on :) | 00:19 |
| bknudson | jamielennox: https://review.openstack.org/#/c/172662/ -- it's not gone yet. | 00:20 |
| fungi | jamielennox: now that python-keystoneclient has an actual stable/juno branch, we shouldn't run that job any longer | 00:20 |
| mordred | lifeless: the current thinking around the semver patches it to move them to a feature branch and make the stable/0.10 branch master again so that we can have a releasable master branch | 00:20 |
| fungi | jamielennox: instead we should just run normal non-backward-compat integration test jobs on all the python-keystoneclient branches | 00:21 |
| SpamapS | I'm really not understanding why any of this is an issue or what we're actually arguing about. | 00:21 |
| tchaypo | from a brief look at https://review.openstack.org/#/c/166703/ I think it’s also running icehouse/juno tests against master | 00:21 |
| mordred | SpamapS: I think we needed to make sure we all agreed | 00:21 |
| SpamapS | Some people don't like requirements.txt seems like the only thing that might matter? | 00:21 |
| lifeless | mordred: please no; has anyone *tried* to do the minor change needed that I laid out in response to doug | 00:21 |
| lifeless | s mail ? | 00:21 |
| *** woodster_ has joined #openstack-infra | 00:22 | |
| mordred | lifeless: I believe at least two of us have looked at it and it was not minor for our brains | 00:22 |
| lifeless | SpamapS: if you don't care, close eyes and walk on bye. If you do have a specific concern with what we resolved, cool, lets discuss it. | 00:22 |
| lifeless | SpamapS: but please don't reopen it in a vague way :) | 00:22 |
| *** yamada-h has quit IRC | 00:22 | |
| lifeless | mordred: ok, then I'll get this autobuilding wheel patch back passing again for dstufft to review, and do the pbr patch. | 00:23 |
| SpamapS | lifeless: I have concern that there's a massive wall of text here with no clear problem statement beyond "We're doing things that upstream pip considers wrong." .. but I may have missed something important. | 00:23 |
| jamielennox | fungi, bknudson: so what is keystoneclient's stable/juno branch then? | 00:23 |
| lifeless | SpamapS: that *is* the problem statement. | 00:23 |
| mordred | lifeless: thanks! that would be much happier making- I don't _Really_ want to effectively revert that branch, but it's been blocked for ages | 00:23 |
| lifeless | SpamapS: the wall of text is examining the consequences. | 00:23 |
| fungi | jamielennox: i don't understand your question. keystoneclient's stable/juno branch is called "stable/juno" | 00:23 |
| mordred | SpamapS: tl;dr - requirements.txt carries a connotation elsewhere that is different than it does here | 00:24 |
| bknudson | jamielennox: if we need to backport a security fix we do it to the stable/juno branch | 00:24 |
| SpamapS | lifeless: ok, and right or wrong requires a test, so is the debate around the content of the test, or the relevance of the test (I see points toward both) | 00:24 |
| bknudson | jamielennox: it's off 1.1.0, so when we fix it's 1.1.1 | 00:24 |
| mordred | SpamapS: I think it's more "we've diverged from the herd, and there is a low-cost way to re-converge" | 00:24 |
| fungi | jamielennox: it's used in exactly the same way as the branch of the same name for all other projects | 00:24 |
| jamielennox | fungi, bknudson: oh, ok. we are back to tagging clients against the OS release | 00:24 |
| jamielennox | i saw that on the ML and didn't realize it was decided | 00:25 |
| bknudson | jamielennox: we decided to cap requirements in stable/branches. | 00:25 |
| lifeless | SpamapS: there is no debate now | 00:25 |
| SpamapS | Ah, ok, so test content clarified (pointer perhaps?) and relevance agreed upon? | 00:25 |
| jamielennox | bknudson: and we're using minor versions for backporting things to stable branches? | 00:25 |
| lifeless | SpamapS: I am so confused. Read the wall of text :). | 00:26 |
| jamielennox | bknudson: ah micro - the third number | 00:26 |
| tchaypo | fungi: when you say " we should just run normal non-backward-compat integration test jobs on all the python-keystoneclient branches” - I think you mean that the normal jobs will notice they’re running on an icehouse branch, so they’ll pull icehouse versions of everything else, so there’s no need to have special jobs for specific branches? | 00:26 |
| bknudson | jamielennox: y, it'll be the micro increases, since there's already a 1.2.0 | 00:26 |
| fungi | jamielennox: yes, because otherwise we can't fix, say, a keystoneclient bug which affects stable/juno servers using it as a means of communicating with one another | 00:26 |
| *** wenlock has quit IRC | 00:26 | |
| fungi | tchaypo: correct, just like they do for all projects with stable branches | 00:26 |
| jamielennox | fungi: yep, understood | 00:26 |
| lifeless | mordred: are you familiar with https://review.openstack.org/#/c/161047/4/specs/stable-requirements.rst ? | 00:26 |
| *** salv-orlando has quit IRC | 00:26 | |
| SpamapS | lifeless: I'll state it more succinctly, sorry. I don't like using right or wrong without a clear idea of what sort of test I need to apply to decide which space anything is in. | 00:27 |
| mordred | lifeless: I did -1 it :) | 00:27 |
| tchaypo | fungi: so in the case of https://review.openstack.org/#/c/166703/ we should be taking those jobs out of python-novaclient as well? | 00:27 |
| lifeless | SpamapS: I don't think right or wrong are useful concepts in the discussion we just had | 00:27 |
| lifeless | SpamapS: I am confused about your use of them here | 00:28 |
| tchaypo | I was only planning to make them non-voting for now, not needing them at all sounds even better | 00:28 |
| SpamapS | lifeless: so if there is a set of rules that upstream or "the herd" has established, that would be helpful to me to understand a) if we're actually doing it wrong, and b) how we can fix it, and finally c) what priority we should commit to fixing it. | 00:28 |
| fungi | tchaypo: if the stable/X branches have been created for novaclient already, then yes | 00:28 |
| lifeless | we're doing it differently. See my mail to openstack-dev for details. And for b). As for c), its tech debt, vs breaking-things-daily. | 00:29 |
| tchaypo | fungi: but also, we should be fixing the pins in global-requirements.txt to use <X.Y+1.0 rather than <=X.Y.Z, right? | 00:29 |
| jamielennox | fungi, bknudson: has the same change been proposed for the other clients? i rechecked https://review.openstack.org/#/c/141994/ but it's essentially failing on the same problem | 00:29 |
| fungi | tchaypo: yes | 00:29 |
| lifeless | mordred: so you and jamespage are in disagreement there. | 00:29 |
| jamielennox | pkg_resources.ContextualVersionConflict: (python-glanceclient 0.17.0.post17 (/opt/stack/new/python-glanceclient), Requirement.parse('python-glanceclient<=0.15.0,>=0.14.0'), set(['ceilometer'])) | 00:29 |
| *** sdake has joined #openstack-infra | 00:29 | |
| tchaypo | fungi: mrmm. no branches for python-novaclient yet. | 00:29 |
| bknudson | jamielennox: I looked at updating all the clients but it failed jenkins :( | 00:30 |
| lifeless | mordred: I think that that spec may well have a useful reason for having a requirements.txt that is like pip expects. | 00:30 |
| fungi | jamielennox: i am unsure if anyone has been incrementally proposing/fixing those when creating the stable branches for the clients | 00:30 |
| lifeless | mordred: all the transitive requirements flattened out and tied down tightly. | 00:30 |
| bknudson | jamielennox: see https://review.openstack.org/#/c/172719/ | 00:30 |
| *** tiswanso has quit IRC | 00:30 | |
| fungi | tchaypo: bknudson: jamielennox: i've not really been involved. i believe ttx has been working with the various ptls on this | 00:30 |
| lifeless | mordred: I don't think we need to reach consensus on that now - but I do think we need to agree that *if* you get convinced on that spec, that the transitive thing wouldn't end up in setup.cfg. | 00:31 |
| bknudson | jamielennox: I think it's best for other projects to propose these themselves since they know what's going on with their libs. | 00:31 |
| SpamapS | lifeless: mmmk. We have a lot of debt, but I think the thing that causes us to break things all the time is probably the most important facing OpenStack deployers (especially CD), so I'll take your word for it that paying this debt down will lead to less instability in that area. | 00:31 |
| bknudson | jamielennox: I think it failed because some of the libs didn't do semver right and have reqs updates in their minor update. | 00:31 |
| lifeless | SpamapS: I've made no such claim | 00:31 |
| jamielennox | bknudson: right, that glanceclient is way outside the restriction even you set and i have no idea what it should be | 00:31 |
| mordred | lifeless: I find it HIGHLY unlikely that I will be convinced that the needs of the gate should be put into a file that pip will ignore | 00:32 |
| bknudson | jamielennox: in the case of keystoneclient we're using 1.1.0, so the cap should be <1.2.0 | 00:32 |
| bknudson | jamielennox: so that we can release a 1.1.1 | 00:32 |
| lifeless | mordred: sure. But if you are convinced, just not setup.cfg. | 00:32 |
| mordred | lifeless: jamespage has also invoked the strawman redherring | 00:32 |
| lifeless | mordred: thats all I want :) | 00:32 |
| mordred | lifeless: his argument is specious and wrong and factually inaccruate | 00:32 |
| *** koolhead17 has joined #openstack-infra | 00:32 | |
| bknudson | jamielennox: glanceclient needs to work the same way, otherwise they can't really do this anyways. | 00:33 |
| jamielennox | bknudson: yea, one of the project (can't remember which) continually bumped the micro only which made it impossible | 00:33 |
| mordred | lifeless: as I explained to tchaypo just a little bit ago | 00:33 |
| mordred | lifeless: the pining of requirements ONLY affects pip installation | 00:33 |
| mordred | and I feel very strongly that we should not ship things that are known to be broken | 00:33 |
| SpamapS | mordred: What about putting the result of pip freeze from the gate -- somewhere that woudl be tied to each commit. ? | 00:33 |
| mordred | SpamapS: I feel very strongly that we should not ship artifacts that are different than what we tested | 00:33 |
| mordred | I think that's very rude | 00:33 |
| tchaypo | fungi: 7bf16e596d3d672b0454cd33243dddaa3b569a26 seems to have quite a few errant <=s in | 00:33 |
| mordred | and highly confusing | 00:34 |
| *** sdake_ has quit IRC | 00:34 | |
| mordred | especially since we have the ability to do the other thing | 00:34 |
| tchaypo | mordred: I think we all agree on that | 00:34 |
| mordred | and it's not hard | 00:34 |
| mordred | tchaypo: woot! | 00:34 |
| lifeless | mordred: so this means you think we should start specifying the transitive requirements in install_requires? | 00:34 |
| lifeless | mordred: I thought you were very much against that? | 00:34 |
| SpamapS | mrodden: Not suggesting that. Suggesting that each commit would ship the freezes that worked, in addition to whatever we tried to use pass the gate. | 00:35 |
| SpamapS | doh | 00:35 |
| fungi | jamielennox: bknudson: there is a workaround where we invalidate the bad micro bumps with != matches in the requirements lists of the appropriate branches, and then tag new non-micro versions far enough apart on those branches | 00:35 |
| SpamapS | mrodden: sorry, mordred ^^ | 00:35 |
| mordred | lifeless: I think that if a transitive depend is so broken taht we cannot pass the gate without pinning it directly in some way | 00:35 |
| mordred | lifeless: then we should put it into install_requires | 00:35 |
| mordred | lifeless: I'm just saying - I do not want a generated set of pins used in the gate _different_ than what a user would get from pip install | 00:36 |
| mordred | if we need to pin in the gate, we need to communicate that we know that to the user | 00:36 |
| mordred | and we need to make the communication of that eh default thing they get | 00:36 |
| mordred | if they want to override it, it's SUPER easy to | 00:36 |
| mordred | since pip does not actually enforce depends | 00:36 |
| lifeless | ah | 00:37 |
| lifeless | it does | 00:37 |
| lifeless | and its getting better at it | 00:37 |
| mordred | neat. pbr also has a SKIP_PIP_INSTALL mode | 00:37 |
| mordred | I'm just sayin | 00:37 |
| jamielennox | fungi: out of interest can't we force the version bump, could the global requirements bot make commit messages that mean you always force at least a minor version bump if you change reqs | 00:37 |
| SpamapS | But it doesn't tell you when you're about to violate an existing package's depends, right? | 00:37 |
| mordred | if the software doesn't work without a pin | 00:37 |
| lifeless | it matters somewhat now, it's going to matter more in future | 00:37 |
| mordred | we should pin | 00:37 |
| mordred | if you feel strongly that you want to override what we've told you is broken | 00:37 |
| mordred | that's on you | 00:37 |
| mordred | and you can figure it out | 00:37 |
| lifeless | so I'm confused again | 00:38 |
| mordred | but the burden should be on the person who thinks they know better tahn the gate | 00:38 |
| lifeless | since you seem to be arguing something that noone proposed | 00:38 |
| SpamapS | Also do we have a sense of how much of "The way upstream works" has to do with how often people deploy in isolation via virtualenv/containers/etc? | 00:38 |
| tchaypo | "I think that if a transitive depend is so broken taht we cannot pass the gate without pinning it directly in some way” - that’s a reactive situation, where we see something *did* break and we need to communicate that users should *not* install that version as it’s known bad | 00:38 |
| mordred | lifeless: have yuo read the sepc? | 00:38 |
| lifeless | AIUI the proposal for the gate is to reduce unexpected churn | 00:38 |
| mordred | lifeless: have you read my -1? | 00:38 |
| mordred | no | 00:38 |
| lifeless | mordred: I have | 00:38 |
| mordred | the proposal is to not do that | 00:38 |
| mordred | it's to have the gate produce a file that is not what is shipped with the software but which is what the gate will pin on | 00:38 |
| tchaypo | but this doesn’t seem to have space for a pro-active notification where we tell users “we expect that all sub-versions of X.Y should work, unless noted otherwise" | 00:38 |
| mordred | and I am categorically opposed to that | 00:38 |
| lifeless | mordred: thats the mechanism, not the reason | 00:39 |
| mordred | lifeless: I am opposed to the mechanism | 00:39 |
| mordred | my -1 says that | 00:39 |
| mordred | I think the proposal in genereal is fine | 00:39 |
| mordred | the mechanism here is highly important | 00:40 |
| mordred | it's the entire subject of our conversation | 00:40 |
| lifeless | mordred: it also says in the summary that the intent is to capture the information so folk can use it later | 00:40 |
| mordred | why would you not capture it in the default instructions for machine installation of the software? | 00:40 |
| mordred | the idea that people should need to opt-in to a good installation is bonghits | 00:40 |
| tchaypo | would you be less opposed if we reversed it - put the broad requirements in requirements.gate, and had the gate generate a requirements.txt that exactly matches what was tested? | 00:41 |
| mordred | why would we have two lists? | 00:41 |
| mordred | what does it help or solve? | 00:41 |
| lifeless | tchaypo: the gate is too late though, the commit has already been made | 00:41 |
| mordred | our requirements should be as broad as they _Can_ be | 00:41 |
| *** amitgandhinz has joined #openstack-infra | 00:42 | |
| mordred | so - publishing the list of what specific versions we used is a good idea | 00:43 |
| mordred | and we do it | 00:43 |
| mordred | already | 00:43 |
| mordred | having a different file that is not used by the installation of te software be used in devstack | 00:43 |
| mordred | leads us to a place where we are shipping softare that is different than what we are testing - which benefits literally no humans | 00:44 |
| *** sdake_ has joined #openstack-infra | 00:46 | |
| tchaypo | lifeless: I was thinking of gating to changes on the global-requirements repo, but I think you’re right - I think you mean we really want something that catches the versions on the repo being tested | 00:47 |
| tchaypo | mordred: where do we publish the list of specific versions that went through CI? | 00:47 |
| bknudson | tchaypo: most of the jobs have the pip freeze output. | 00:48 |
| bknudson | btw - it would be great if we could run the tests with the min versions of all the packages, too. | 00:49 |
| *** sdake has quit IRC | 00:49 | |
| tchaypo | How does someone installing or distributing the package get access to that though? | 00:51 |
| *** koolhead_ has joined #openstack-infra | 00:52 | |
| *** sdake has joined #openstack-infra | 00:52 | |
| bknudson | we don't... we look at requirements.txt files | 00:52 |
| tchaypo | If I was to do it, I think I’d have to figure out the last check on the last review that landed before the release, then grovel through the output | 00:52 |
| tchaypo | To me that doesn’t sound like something we “published” - especially since iirc the logs for the gate checks get rotated out after a few months | 00:53 |
| mordred | tchaypo: right. that's why I have been advocating hard-pins of all our stable braches for several years now | 00:54 |
| bknudson | probably the way to do it is to pin the stable/ requirements.txt on release. | 00:54 |
| mordred | bknudson: ++ | 00:54 |
| mordred | the only argument agains that is "some upstream python library might release a bugfix" | 00:55 |
| mordred | which has categorically been proven nothing but untrue | 00:55 |
| fungi | bknudson: the trick is identifying "minimum versions of our requirements" | 00:55 |
| *** koolhead17 has quit IRC | 00:55 | |
| fungi | bknudson: we have a tool which can install "the highest version matching a given set of constraints" and that tool is called pip | 00:55 |
| mordred | that this is so complex should make it clear that anything other than hard-pins on stable is crazy | 00:55 |
| lifeless | ok, commented on the spec | 00:55 |
| mordred | why do we care about lowest versions? | 00:56 |
| mordred | it is not useful | 00:56 |
| *** sdake_ has quit IRC | 00:56 | |
| mordred | the only thing that is useful is "_this_ will install and work" | 00:56 |
| mordred | if other people want to look at other combinations of stuff - awesome | 00:56 |
| fungi | i agree. the only reason i can see for testing declared minimums is to see of they're actually legitimate | 00:56 |
| lifeless | mordred: so the compat thing is that we have a dozen projects | 00:56 |
| lifeless | they all need to be mutually installable today | 00:57 |
| bknudson | I don't think the current mins are legitimate. | 00:57 |
| fungi | i'm quite sure they're not | 00:57 |
| bknudson | at least the keystoneclient tests don't work. | 00:57 |
| mordred | lifeless: yup | 00:57 |
| tchaypo | and we have downstream people preparing distributions of openstack+other things, and we need openstack to be mutually installable with those things | 00:57 |
| mordred | lifeless: that has been echoed by the operators, btw, not just the distros | 00:57 |
| lifeless | thats why install_requires *must* be as broad as it can | 00:57 |
| mordred | I disagree | 00:57 |
| mordred | that's why install_requires must be as strict as possible | 00:57 |
| lifeless | we can't change a dozen projects atomically | 00:57 |
| mordred | as strict _As_ _possible_ | 00:57 |
| fungi | tchaypo: we also publish a robust testsuite which downstreams can use to find out of the versions of things they already have packaged work with what we're releasing, without us needing to check that for them | 00:58 |
| fungi | er, find out if | 01:01 |
| tchaypo | and if a downstream wants to start their testing by reproducing exactly what we tested - how do they do that? | 01:01 |
| lifeless | mordred: we get that strictness wrong a lot IME | 01:02 |
| lifeless | mordred: such as failing to bump lower versions during dev cycles | 01:02 |
| lifeless | mordred: before using features. | 01:03 |
| lifeless | mordred: I don't know what the answer is here. | 01:03 |
| dstufft | mordred: it's not just bug fixes, if you're strict you make it hard (or impossible) to actually isntall your thing | 01:03 |
| dstufft | alongside other things I mean | 01:03 |
| dstufft | you're basically promising dependency hell unless you stick every single top level thing in it's own virtualenv | 01:04 |
| *** Longgeek has joined #openstack-infra | 01:04 | |
| jamielennox | bknudson: with your change or after? | 01:06 |
| mordred | dstufft: I do not believe we're suggesting that installing our thing alongside somethign else is a feasible thing | 01:07 |
| fungi | tchaypo: they can look at the pip freeze output from our tests, if they're trying to reproduce an explicit set of our test results. that's why we include pip freeze output in the logs | 01:07 |
| bknudson | jamielennox: what's the question about? | 01:07 |
| dstufft | mordred: what's the "thing" in question? For the python clients it's absolutely a feasible thing | 01:07 |
| dstufft | if it's not you might as well jsut rm -rf them | 01:07 |
| *** baoli has quit IRC | 01:07 | |
| mordred | dstufft: the python clients purpose in life is intra-service communication | 01:07 |
| jamielennox | bknudson: oh, mentioning ksc highlighted your last message, didn't realize it wasn't directed at me - nevermind | 01:07 |
| mordred | dstufft: they are not written to be use facing | 01:08 |
| mordred | user | 01:08 |
| mordred | dstufft: so, yes, you might as well rm -rf them | 01:08 |
| mordred | they are a part of the servers - and outside of that you probably need to install them into venvs | 01:08 |
| bknudson | jamielennox: see https://review.openstack.org/#/c/172655/ | 01:08 |
| bknudson | jamielennox: we were talking about that the other day... I tried using mock before. | 01:08 |
| dstufft | mordred: so nobody else is supposed to depend on them or use them? | 01:08 |
| mordred | dstufft: we're talking about stable branches here | 01:09 |
| mordred | so no | 01:09 |
| mordred | end users should use the latest release of a client lib | 01:09 |
| mordred | which should not be aggressively pinned | 01:09 |
| mordred | because it is the thing that will work with everything | 01:09 |
| jamielennox | bknudson: yep, i remember - nice that it doesn't use mock, scary that we have code that doesn't care what the cookie is, just that something has been set | 01:09 |
| lifeless | mordred: now, I think we're making progress. I've realised a source of cognitive dissonance. | 01:10 |
| fungi | dstufft: server x imports a "client library" corresponding to server y and then uses it to communicate with server y | 01:10 |
| lifeless | mordred: see the list :) | 01:10 |
| mordred | however, with a juno release, what nova needs to know is that the copy of keystoneclient it installs on its bo will work | 01:10 |
| mordred | lifeless: looking | 01:10 |
| dstufft | mordred: maybe I'm confused, I thought that python-novaclient was like boto for AWS, except just for nova, where you'd use it in your own application | 01:10 |
| lifeless | dstufft: we thought that for ages | 01:11 |
| lifeless | dstufft: and some of the projects do think of their library like that. | 01:11 |
| fungi | dstufft: so the version of the library that existed at the time those servers were released should be able to work with that release version of those servers as far as their needs for communicating with one another is concerned | 01:11 |
| lifeless | dstufft: but | 01:11 |
| lifeless | dstufft: tere is now a specific UI one - openstackclient | 01:11 |
| mordred | lifeless: my main concerns is that python libraries all suck | 01:11 |
| dstufft | s/python // | 01:11 |
| mordred | dstufft: ++ | 01:11 |
| lifeless | mordred: yes they do | 01:11 |
| mordred | lifeless: and they break things CONSTANTLY | 01:11 |
| lifeless | mordred: and you are right about that | 01:11 |
| lifeless | mordred: BUT | 01:12 |
| mordred | lifeless: so, what I want to make sure we don't do | 01:12 |
| mordred | for our _stable_ branches | 01:12 |
| mordred | is test one thing and ship another | 01:12 |
| mordred | in the name of "flexibility" | 01:12 |
| lifeless | mordred: flexability isn't the cause here | 01:12 |
| lifeless | mordred: I can state the problem without that strawman - I think I did in fact | 01:12 |
| mordred | lifeless: I would like you to try again, because I don't think you did :) | 01:12 |
| fungi | dstufft: essentially, if debian packages nova and glance, and nova is using python-glanceclient to communicate with glance, the version of python-glanceclient they'll have packaged will be comtemporary with the versions of nova and glance they packaged. we're doing similar things on our stable branches of python-glanceclient to support that use case (as opposed to the end-user/app-developer needs to | 01:13 |
| fungi | communicate with a cloud service use case) | 01:13 |
| dstufft | fungi: so my point was really that if you expect random end users who want to interact with an openstack cloud to install these things, you can't use == inside of python-*client (or any of their dependencies) | 01:14 |
| mordred | dstufft: right. but that's the thing | 01:14 |
| mordred | dstufft: we cannot use that in the python-*clent we expect users to install | 01:14 |
| mordred | dstufft: 100% agree | 01:14 |
| lifeless | mordred: https://etherpad.openstack.org/p/stable-omg-deps | 01:14 |
| tchaypo | fungi: *thinks* for someone looking at the source, that’s probably easier than for someone getting the package - they can at least look at the git log, find the last commit sha, look that up on gerrit, find the last set of gate checks on the review.. | 01:15 |
| fungi | dstufft: this is the use case of the library being installed onto a server where these services are also running, for communicating with one another | 01:15 |
| mordred | dstufft: pinned versions of client libs are not expected to be installed by users | 01:15 |
| *** unicell has quit IRC | 01:15 | |
| mordred | dstufft: I am 100% against them doing that | 01:15 |
| tchaypo | fungi: but am I wrong in thinking that the pip-freeze output is eventually going to go away as the logs get rotated out? | 01:15 |
| dstufft | mordred: does openstack support mix and match versions of the server libraries? | 01:15 |
| dstufft | s/libraries/processes/ | 01:15 |
| lifeless | mordred: thats a problem statement | 01:15 |
| dstufft | like can you have nova from version X and swift from version Y | 01:16 |
| lifeless | mordred: based on what I believe you want us to achieve | 01:16 |
| lifeless | dstufft: ish | 01:16 |
| fungi | the servers are unfortunately too tightly-coupled to one another and to their use of those libraries for intercommunication, such that they can't well replace the version which was available when they released with a much newer version of the library and still talk to each other, so we need to have multiple release branches for the libraries | 01:16 |
| lifeless | dstufft: swift releases outside the 6-month cycle | 01:16 |
| dstufft | lifeless: well I just pciked a random two names that first poped into my head | 01:16 |
| lifeless | dstufft: so some skew is expected. The rest of the project only test skew during upgrades, not for regular use | 01:16 |
| mordred | dstufft: nova needs to know which version of the python library it is going to use | 01:16 |
| dstufft | replace swift with glance or wahtever | 01:16 |
| mordred | dstufft: that version of the client libary will support more than just that version of other servers it's going to talk to | 01:16 |
| dstufft | == would mandate that you upgrade everything in lock step if you install two servers into the same environment of different versions | 01:17 |
| mordred | dstufft: so, the client libraries have a decent range of support for remote rest protocol | 01:17 |
| *** yamahata has joined #openstack-infra | 01:17 | |
| *** unicell has joined #openstack-infra | 01:17 | |
| lifeless | fungi: we could try to actually fix that coupling :) Or bundle all our things into one release rather than 4 billion permutations. But thats a different near-rant. | 01:17 |
| mordred | dstufft: but if nova stable/juno depends on python-novaclient==2.12.20 - that's about the _python_ interface it needs | 01:17 |
| lifeless | mordred: did you do the ++ in the etherpad ? | 01:17 |
| mordred | lifeless: yes | 01:17 |
| fungi | lifeless: i think we _should_ fix that, but i doubt we could get all the projects to just stop doing anything else until we get there | 01:17 |
| lifeless | fungi: if we didn't work around that... anyhow. tangent. | 01:18 |
| dstufft | mordred: sure, buit what you want to also install barbican unstable/wahtever which maybe depends on python-novaclient==2.15? | 01:18 |
| lifeless | ok so we have a problem statement | 01:18 |
| mordred | dstufft: you can't do that | 01:18 |
| mordred | dstufft: it's completely unsupported and always will be | 01:18 |
| dstufft | mordred: ore ven beyond that, what about a 2nd release from the stable branch with an updated requirement | 01:18 |
| fungi | you get to keep both pieces, as they say | 01:18 |
| dstufft | so if you have 2015.4.0 and 2015.4.1 | 01:19 |
| mordred | dstufft: that 2nd release from the stable branch will need to correspond with making sure that nova also has a release | 01:19 |
| *** yamada-h has joined #openstack-infra | 01:19 | |
| mordred | dstufft: or that the release is compatable with the stable nova that exists | 01:19 |
| *** dimsum_ has quit IRC | 01:19 | |
| mordred | dstufft: doing that is tractable and we can make sure it works | 01:19 |
| dstufft | mordred: right, but you're going to force deployers to upgrade _everything_ in lock step in that scenario instead of being able to do it one server process at a time | 01:19 |
| mordred | dstufft: including unstable barbican on the same box installed via pip without virtualenvs is not | 01:19 |
| mordred | dstufft: most deployers arent' installing via pip | 01:19 |
| mordred | dstufft: most are making their own distro packages | 01:20 |
| mordred | dstufft: the ones who are installing via pip are insatlling into virtualenvs and/or containers | 01:20 |
| fungi | in general the x.y.* versions for one project are expected to work with the a.b.* versions of another project which released in the same timeframe and which use the same branch names | 01:20 |
| lifeless | I'm now brainstorming solutions on that etherpad. | 01:21 |
| tchaypo | lifeless: mordred: I understand that rpm/deb installers don’t use pip; but my understanding is that the distro toolchains will use pip’s idea of what needs to be installed as the basis for the OS package’s list of dependencies | 01:21 |
| *** achanda has joined #openstack-infra | 01:21 | |
| *** baoli has joined #openstack-infra | 01:21 | |
| dstufft | mordred: btw, how do you even test this? Doesn't this require that you can merge a patch to update the pinned version across all projects at once? You can't land it in one, run it through CI, thne land it in another | 01:22 |
| tchaypo | if our list is too tight, it requires manual work from the maintainers to find something that fits with the rest of the release; if we can be looser (but still tight enough that they have confidence about what’s tested and expected to work) it reduces the manual work they have to do | 01:22 |
| tchaypo | but I have never been an upstream distro maintainer so I could be very wrong, again.. | 01:22 |
| fungi | tchaypo: i think the distros are going to want to use the versions of things they've already packaged if possible, plus the newest versions of things we claim to support which they haven't yet packaged. seems like any other expectation is pathological | 01:22 |
| tchaypo | fungi: sounds reasonable | 01:23 |
| *** yamada-h has quit IRC | 01:23 | |
| fungi | we also provide tools they can use to test those things together and double-check whether they'll work | 01:23 |
| mordred | dstufft: we do an override sync with teh global requirements in our integration tests | 01:23 |
| dstufft | mordred: do you do that in a way that it gets piped into isntall_requires? | 01:24 |
| mordred | dstufft: yes | 01:24 |
| dstufft | ok | 01:24 |
| *** liusheng has quit IRC | 01:24 | |
| *** amotoki has joined #openstack-infra | 01:25 | |
| *** liusheng has joined #openstack-infra | 01:28 | |
| *** dimsum_ has joined #openstack-infra | 01:28 | |
| tchaypo | does that mean that the requirements.txt in individual project repos actually don’t represent the input to the CI system? | 01:30 |
| *** mriedem has quit IRC | 01:30 | |
| *** mriedem has joined #openstack-infra | 01:31 | |
| lifeless | dstufft: but we don't have to. We could equally use a separate file. We *don't capture the override contents today*' - at least AFAIK | 01:31 |
| lifeless | I need to grab food. You folk not on etherpad should peek :). | 01:34 |
| *** baoli has quit IRC | 01:35 | |
| *** dimsum_ has quit IRC | 01:35 | |
| *** otter768 has joined #openstack-infra | 01:35 | |
| * tchaypo hunts for etherpad link | 01:36 | |
| fungi | tchaypo: they're used in most single-repo jobs, e.g. for unit tests, but devstack-gate integration testing jobs specifically have a routine which overwrites them with what's in the corresponding openstack/requirements branch so that it has a consistent set when co-installing all of them | 01:36 |
| mordred | fungi, tchaypo: just had some fairly complete thoughts about test metadata annotation post-commit into git notes on openstack/openstack | 01:41 |
| fungi | we sort of capture the requirements used, insofar as we checkout a specific ref from the openstack/requirements repo in an integration test and that is logged | 01:41 |
| tchaypo | so both sets of requirements count as “tested in ci” because they’re tested in different checks? | 01:41 |
| fungi | right | 01:42 |
| fungi | we would like to always test what is actually declared in individual repos, but for the sake of single-env integration between multiple repos that's just not tractable, hence the rewriting we perform | 01:43 |
| mordred | fungi, tchaypo, jeblair (when around): https://etherpad.openstack.org/p/stable-omg-deps - starting aroudn line 27 | 01:43 |
| mordred | we could record dpkg-get-selections that way too | 01:44 |
| mordred | so that deployers and folks could dig back through and find out exactly what was going on | 01:44 |
| mordred | ok. with that -I need to eat some food | 01:45 |
| tchaypo | it’s 11:45am, so I should probably get that breakfast I’ve been putting off as well | 01:45 |
| tchaypo | Do we want pip installs of stable releases to be possible as well? | 01:46 |
| tchaypo | I’ll speculatlively add something and we can remove it if it’s wrong | 01:46 |
| mordred | tchaypo: I think so, yes | 01:46 |
| clarkb | whats the issue here? | 01:49 |
| clarkb | the problem statement is false | 01:51 |
| clarkb | you CAN reproduce based on the freeze capture | 01:51 |
| tchaypo | clarkb: if you can find the freeze capture. | 01:52 |
| clarkb | its in every job log... | 01:52 |
| tchaypo | if that log hasn’t been wiped | 01:52 |
| tchaypo | I just did “pip install nova” and I expect pip to reproduce what CI did - how should pip go about finding the freeze capture that’s related to the package I’m installing? | 01:52 |
| clarkb | how is that different than anything else? | 01:53 |
| clarkb | yes if you delete the data its gone | 01:53 |
| tchaypo | which changes your statement from “you CAN reproduce” to “you MAY be able to reproduce" | 01:53 |
| clarkb | tcgaypo you go get it | 01:53 |
| clarkb | no its just as CAN as any other option | 01:53 |
| tchaypo | I’d like it to be more CAN | 01:53 |
| lifeless | tchaypo: pip install nova can't reproduce what CI did for two reasons. | 01:54 |
| clarkb | I thibk my concern is that we will spend so much time fixing a broken pip system with more hacks | 01:54 |
| clarkb | when we really need to just fix pip | 01:54 |
| lifeless | clarkb: we are fixing pip | 01:54 |
| lifeless | clarkb: I'm trying to get rid of our local hacks. Its what provoked this discussion at all. | 01:55 |
| lifeless | clarkb: mordred has been focused on making our install_requires provide as close a match to what CI tested as possible. | 01:55 |
| tchaypo | +1 for fixing the problem in pip if possible | 01:55 |
| tchaypo | but I think part of that is going to be giving pip better data to use | 01:55 |
| lifeless | clarkb: this doesn't work - we know it doesn't work because we don't actually use our install_requires at all anyway. We override it! | 01:56 |
| clarkb | lifeless I do not see make pip resolve deps properly on the etherpad | 01:56 |
| clarkb | anything else is not fixing pip | 01:56 |
| lifeless | clarkb: it doesn't need to give better reproducability. | 01:56 |
| clarkb | it does | 01:56 |
| clarkb | because install resolution can change over time | 01:56 |
| clarkb | because it doesnt resolve sanely | 01:56 |
| lifeless | clarkb: no, it doesn't. https://github.com/pypa/pip/issues/988 1a. | 01:56 |
| lifeless | clarkb: a requirements.txt file (*not* what pbr does) will win, assuming everything else is compatible. | 01:57 |
| lifeless | clarkb: which it will be, because it was at the point in time CI ran. | 01:57 |
| clarkb | order matters here and its depth first iirc | 01:57 |
| clarkb | requirements.txt doesnt always win today | 01:58 |
| lifeless | clarkb: order matters, which is why everything will be there. | 01:58 |
| lifeless | clarkb: what happens today is pbr, so install_requires, *not* -r. | 01:58 |
| clarkb | but order can change transitively | 01:58 |
| lifeless | no, it can't. | 01:58 |
| clarkb | why not? | 01:58 |
| *** mriedem has quit IRC | 01:58 | |
| lifeless | because we're giving pip everything up front. All the actual discovery is going to happen after that. | 01:58 |
| clarkb | you are going to list every transitive dep? how do you do that? | 01:58 |
| lifeless | pip freeze | 01:59 |
| clarkb | ok so the pypi mirror hack | 01:59 |
| lifeless | huh? | 01:59 |
| clarkb | install everything | 01:59 |
| clarkb | then check | 01:59 |
| lifeless | That seems like a non sequitor | 01:59 |
| clarkb | then do it again | 01:59 |
| lifeless | no | 01:59 |
| lifeless | we do CI | 01:59 |
| fungi | e.g. give up on using pip like a package manager which can find transitive dependencies and just use it as a "install this list of things i don't care of they depend on other things i worked it out for you already" | 01:59 |
| lifeless | using what we do today. | 01:59 |
| lifeless | And we capture the end result for other folk to use to reproduce what we used. | 02:00 |
| lifeless | Thats the problem statement. | 02:00 |
| clarkb | we already do that... | 02:00 |
| lifeless | fungi: for the specific case of 'reproduce what CI did' | 02:00 |
| clarkb | I really dint understand what needs to change. maybe where the pip freeze is stored? | 02:00 |
| fungi | oh, and then have a different list for people who just want to `pip install python-novaclient` or whatever? | 02:00 |
| clarkb | also phone keyboard is terribad | 02:00 |
| lifeless | fungi: right. pip install python-novaclient does not need to reproduce CI exactly. It can't ever. | 02:01 |
| fungi | yeah, typing from bed is not much better | 02:01 |
| lifeless | fungi: so we decouple the problem, and do a near-perfect job for reproducing what CI had - including dpkg package lists and so on. | 02:01 |
| tchaypo | I feel like we need to remind a bunch of you that “This is sunday, you should be doing something fun" | 02:02 |
| dstufft | reproducing an enivornment is basically what requirements is for | 02:02 |
| fungi | lifeless: so i think part of "make a simple way to reproduce _the_one_environment_ in which this commit was tested" presupposes that we first solve making all jobs which test that commit use an identical environment? | 02:02 |
| lifeless | fungi: and let pip install python-novaclient do what its designed to do, which is to exclude everything known not to work | 02:02 |
| tchaypo | except that you might actually think this is fun | 02:02 |
| dstufft | ora t least one of the things | 02:02 |
| fungi | or do you expect to have multiples of these recorded? | 02:02 |
| lifeless | fungi: I'd start by recording one per environment - name it after the job | 02:02 |
| fungi | k | 02:03 |
| clarkb | lifeless we literally already do that | 02:03 |
| clarkb | its one per env based on the job | 02:03 |
| fungi | well, except for multi-virtualenv jobs | 02:03 |
| fungi | e.g. tempest | 02:03 |
| *** weshay has quit IRC | 02:03 | |
| lifeless | clarkb: we don't record it persistently. | 02:04 |
| fungi | that seems like a different problem which we can tackle easily enough | 02:05 |
| fungi | record that to a different file with a different retention period | 02:05 |
| lifeless | clarkb: we also don't tie it back to the commit very well | 02:05 |
| fungi | some additional indexing can help there? | 02:05 |
| lifeless | clarkb: I think both of those things need to be solved to make this as convenient as I think it needs to be for mordred to relax his death grip on 'pip install' | 02:05 |
| lifeless | we have 1000 developers | 02:05 |
| clarkb | lifeless sure we can make it better | 02:06 |
| clarkb | I just dont understand what we are solving | 02:06 |
| clarkb | from what I read it is already done | 02:06 |
| lifeless | I'll let you and mordred debate that | 02:06 |
| fungi | lots of different somewhat-interrelated things from what i can gather | 02:06 |
| clarkb | we have apt/yum listings and pip freezes | 02:06 |
| lifeless | If it was already 'done', I don't think he'd care nearly so much about install_requires being reconstruction of CI | 02:06 |
| lifeless | clarkb: if you're referring to whats in the job logs... there's an awful lot of knowledge needed to use that to make nova head work. | 02:07 |
| clarkb | yes it is a job log | 02:07 |
| clarkb | we can publish a known set somewhere | 02:08 |
| clarkb | but the source is always a job log because thats what it is | 02:08 |
| fungi | maybe going up another level would help. what sort of command would you want to run, and what sort of specific outcome would you want from that command? that would be a start toward being able to design the tools and data sources to support it | 02:08 |
| lifeless | right | 02:09 |
| *** koolhead17 has joined #openstack-infra | 02:09 | |
| tchaypo | I’d like for the published set to be stored close to the package/repo that’s going to be used to do the install | 02:09 |
| lifeless | mordreds example is 'pip install git+..../nova.git' | 02:09 |
| tchaypo | I’d like to avoid getting in a situation where we still have the source but can’t reproduce the install because $WEB_SERVICE is down, or lost data | 02:09 |
| lifeless | which I do not believe can ever work that literally, because of incompatible constraints | 02:10 |
| clarkb | lifeless there is a chickenand egg there | 02:10 |
| clarkb | cant have a commit with known deps until it merges | 02:10 |
| clarkb | so yes | 02:10 |
| lifeless | I think we could sensibly do 'bindep install git+..../openstack/openstack.git/nova';'pip install -r git+.../openstack/openstack.git/nova-requirements.txt' or something like that | 02:11 |
| *** sdake_ has joined #openstack-infra | 02:11 | |
| fungi | lifeless: and he would want `pip install git+..../nova.git` to specifically install the nova dependencies with which that particular commit was tested (forgetting for a moment that there are different dependency sets used for different test jobs)? | 02:11 |
| *** koolhead_ has quit IRC | 02:11 | |
| lifeless | fungi: "On 13 April 2015 at 12:53, Monty Taylor <mordred@inaugust.com> wrote: | 02:12 |
| lifeless | > What we have in the gate is the thing that produces the artifacts that | 02:12 |
| lifeless | > someone installing using the pip tool would get. Shipping anything with | 02:12 |
| lifeless | > those artifacts other that a direct communication of what we tested is | 02:12 |
| lifeless | > just mean to our end users. | 02:12 |
| lifeless | " | 02:12 |
| fungi | okay, so specifically recreating the environment in which python setup.py sdist was run in the tarball/wheel jobs | 02:12 |
| *** yamamoto has joined #openstack-infra | 02:13 | |
| lifeless | fungi: I read that differently. I read it as recreating the environment in which tempest passed. | 02:14 |
| fungi | as those are the only artifacts i can think of our "gate" (really it's not happening in the "gate" pipeline, but whatever) produces which people then install | 02:14 |
| lifeless | fungi: as far as dependency and code goes, not configuration. | 02:14 |
| *** sdake has quit IRC | 02:14 | |
| lifeless | fungi: yes, the language was terrible. The intent was, I think, clear. Users should get what we know works. | 02:14 |
| clarkb | 1) you need to figure out which set(s) your users want 2) if sets disagree what is resolution? | 02:15 |
| fungi | hrm. i still think we're a long way from having "a consistent thing we know works" that we could even communicate to them | 02:15 |
| *** salv-orlando has joined #openstack-infra | 02:15 | |
| lifeless | sets? | 02:15 |
| clarkb | right I think ^ is problem to solve first | 02:15 |
| clarkb | then worry about publishing | 02:15 |
| clarkb | lifeless we test on at least 2 operating systens with multiple python versions and mutliple input requirement seeds | 02:16 |
| fungi | what "works" is currently a nebulous set of different kinds of tests run in all sorts of different kinds of environments | 02:16 |
| clarkb | this all produces differebt outputs that are known to work | 02:16 |
| lifeless | clarkb: yes, thus my proposing we capture each | 02:16 |
| clarkb | lifless but if I pip install locally I only want 1 | 02:17 |
| clarkb | so we need to filter/distill the one thing | 02:17 |
| fungi | step 1: capture information on all our different job environments, step 2: ..., step 3: profit! | 02:17 |
| tchaypo | I think we also have multiple sets of users - deployers vs distributors at least? | 02:17 |
| lifeless | clarkb: huh, the example I gave - a requirements.txt file - lets you choose. | 02:17 |
| tchaypo | but probably they can use much the same set of information for their purposes | 02:17 |
| lifeless | clarkb: the example of pip install git+.../nova.git is problematic, but thats yet another reason trying to cram it into install_requires won't work. | 02:18 |
| openstackgerrit | Joshua Hesketh proposed openstack-infra/os-loganalyze: Tidy up generators into contained objects https://review.openstack.org/172711 | 02:18 |
| openstackgerrit | Joshua Hesketh proposed openstack-infra/os-loganalyze: Add support for non-text files https://review.openstack.org/107267 | 02:18 |
| *** baoli has joined #openstack-infra | 02:18 | |
| fungi | right, as you say we could make is so they could choose to install the environment in which tempest was known to pass, or the environment in which nova python 2.7 unit tests were known to pass, or ... or ... | 02:18 |
| *** salv-orl_ has joined #openstack-infra | 02:18 | |
| lifeless | I don't think unit tests are a candidate here. But if its no harder to capture that, sure. | 02:19 |
| lifeless | capture == massage into easy to use fashion. | 02:19 |
| *** yamada-h has joined #openstack-infra | 02:19 | |
| fungi | though tempest is still a bad (or perhaps important?) example because it uses multiple environments to run one job (so that it can potentially install multiple versions of the same thing on one server) | 02:19 |
| *** yamahata has quit IRC | 02:20 | |
| *** salv-orlando has quit IRC | 02:20 | |
| lifeless | clarkb: how did you go with ovs? | 02:21 |
| *** zz_dimtruck is now known as dimtruck | 02:21 | |
| clarkb | I am not sure that dumping all that info is any more helpful than the info dumo we do today | 02:21 |
| clarkb | lifeless terrible | 02:21 |
| lifeless | clarkb: now is clearly not the time. Tomorrow I can help you if you like. After C's swimming lesson in the morning, that is. | 02:21 |
| clarkb | it doesnt work, fungi and markmcclain were going to lend tgeir eyeballs too | 02:21 |
| clarkb | lifeless sure ping me when you have time | 02:21 |
| fungi | i haven't made any modifications to those held workers yet | 02:22 |
| clarkb | ok I can hold more too | 02:22 |
| *** yamada-h has quit IRC | 02:24 | |
| clarkb | I wonder if it would be easier to start with much more control of the input env | 02:25 |
| clarkb | and just advertise what that is ( though with pip likely not ) | 02:25 |
| lifeless | clarkb: how do you mean? | 02:27 |
| clarkb | have a single versioned env | 02:27 |
| clarkb | so we tested on foodistro version X | 02:27 |
| clarkb | then say 2016.1.1 is compat with ^ | 02:28 |
| tchaypo | so that would eliminate things like differences in underlying C libraries between ubuntu and fedora, for instance? | 02:28 |
| clarkb | right | 02:28 |
| clarkb | not surr it helps the pip install case though | 02:29 |
| lifeless | its an aspect of the same issue | 02:29 |
| tchaypo | well, not eliminiate the differences - but eliminiate the problem where a given set of our packages are tested/working on fedora but broken on ubuntu | 02:29 |
| fungi | yeah, i'm still a little iffy on, for example, bindep caring about package versions explicitly. it's _rare_ that someone has multiple versions of a single distro package name available to them on a particular distro release | 02:32 |
| fungi | saying "if you're running ubuntu 14.04.* install this package name" makes sense | 02:33 |
| fungi | including the versions of distro packages present on the system in what we list for this purpose is of course fine, since it's useful for comparing environments and trying to work out if a bug was introduced in a distro backport patch or something | 02:34 |
| fungi | but having bindep magically figure out how to get that exact version of a distro package seems like a lot of work for very little gain | 02:34 |
| fungi | anyway, i'm out for the night. later all | 02:35 |
| *** dimsum_ has joined #openstack-infra | 02:35 | |
| *** Longgeek has quit IRC | 02:38 | |
| *** ricolin has quit IRC | 02:39 | |
| *** Longgeek has joined #openstack-infra | 02:40 | |
| *** dimsum_ has quit IRC | 02:41 | |
| *** achanda has quit IRC | 02:44 | |
| *** achanda has joined #openstack-infra | 02:48 | |
| *** salv-orl_ has quit IRC | 02:49 | |
| *** oomichi has quit IRC | 02:55 | |
| *** amitgandhinz has quit IRC | 02:55 | |
| openstackgerrit | Merged openstack-infra/project-config: Make a check experiemtal bandit for for python-magnumclient https://review.openstack.org/172460 | 02:56 |
| *** achanda has quit IRC | 02:59 | |
| *** amitgandhinz has joined #openstack-infra | 02:59 | |
| *** pal has joined #openstack-infra | 03:00 | |
| openstackgerrit | Merged openstack-infra/project-config: Fix Python 3 issues https://review.openstack.org/172239 | 03:08 |
| *** subscope_ has joined #openstack-infra | 03:12 | |
| *** amitgandhinz has quit IRC | 03:12 | |
| *** baoli has quit IRC | 03:15 | |
| *** dkehn has joined #openstack-infra | 03:18 | |
| *** woodster_ has quit IRC | 03:20 | |
| *** dhritishikhar has joined #openstack-infra | 03:22 | |
| *** asettle is now known as asettle-lunch | 03:22 | |
| *** yamamoto has quit IRC | 03:23 | |
| *** yamamoto has joined #openstack-infra | 03:23 | |
| *** koolhead17 has quit IRC | 03:24 | |
| *** dkehn has quit IRC | 03:25 | |
| *** sdake_ has quit IRC | 03:25 | |
| *** unicell1 has joined #openstack-infra | 03:35 | |
| *** unicell has quit IRC | 03:37 | |
| *** salv-orlando has joined #openstack-infra | 03:39 | |
| *** woodster_ has joined #openstack-infra | 03:47 | |
| *** gpocentek has joined #openstack-infra | 03:48 | |
| *** gpocentek has quit IRC | 03:48 | |
| *** gpocentek has joined #openstack-infra | 03:48 | |
| *** ajmiller has quit IRC | 03:50 | |
| *** dhritishikhar has quit IRC | 03:51 | |
| *** otter768 has quit IRC | 03:54 | |
| *** crc32 has joined #openstack-infra | 03:54 | |
| *** amitgandhinz has joined #openstack-infra | 03:56 | |
| *** dhritishikhar has joined #openstack-infra | 03:57 | |
| *** crc32 has quit IRC | 03:58 | |
| *** achanda has joined #openstack-infra | 03:59 | |
| *** amitgandhinz has quit IRC | 04:01 | |
| *** salv-orlando has quit IRC | 04:02 | |
| *** achanda has quit IRC | 04:05 | |
| *** asettle-lunch is now known as asettle | 04:08 | |
| openstackgerrit | Merged openstack-infra/project-config: Add check-tempest-dsvm-f21 to experimental queue for openstack/nova https://review.openstack.org/171795 | 04:10 |
| *** unicell1 has quit IRC | 04:11 | |
| *** unicell has joined #openstack-infra | 04:12 | |
| openstackgerrit | Merged openstack-infra/project-config: keystone bandit job on check https://review.openstack.org/170569 | 04:13 |
| openstackgerrit | Merged openstack-infra/project-config: Make a check experimental bandit job for Magnum https://review.openstack.org/171954 | 04:15 |
| *** rm_work|away is now known as rm_work | 04:17 | |
| *** koolhead17 has joined #openstack-infra | 04:19 | |
| *** yamada-h has joined #openstack-infra | 04:21 | |
| *** yamada-h has quit IRC | 04:26 | |
| *** kaisers has joined #openstack-infra | 04:27 | |
| *** BharatK has joined #openstack-infra | 04:29 | |
| openstackgerrit | Merged openstack-infra/project-config: Add a job to build Gerrit javamelody plugin ver 2.10 https://review.openstack.org/172692 | 04:30 |
| openstackgerrit | Merged openstack-infra/project-config: Publish shade docs to docs.openstack.org/infra https://review.openstack.org/170223 | 04:35 |
| *** tiswanso has joined #openstack-infra | 04:35 | |
| *** mrmartin has joined #openstack-infra | 04:37 | |
| *** deepakcs has joined #openstack-infra | 04:39 | |
| *** subscope_ has quit IRC | 04:43 | |
| *** tiswanso has quit IRC | 04:49 | |
| *** amitgandhinz has joined #openstack-infra | 04:57 | |
| *** sabeen has quit IRC | 05:04 | |
| *** liusheng has quit IRC | 05:06 | |
| *** liusheng has joined #openstack-infra | 05:06 | |
| *** yfried has quit IRC | 05:12 | |
| *** sks has joined #openstack-infra | 05:14 | |
| *** btully has joined #openstack-infra | 05:18 | |
| *** yamada-h has joined #openstack-infra | 05:22 | |
| *** dimtruck is now known as zz_dimtruck | 05:24 | |
| *** yamada-h has quit IRC | 05:26 | |
| *** dhritishikhar has quit IRC | 05:46 | |
| *** koolhead17 has quit IRC | 05:46 | |
| *** dhritishikhar has joined #openstack-infra | 05:48 | |
| *** ildikov has quit IRC | 05:53 | |
| *** achanda has joined #openstack-infra | 05:53 | |
| *** otter768 has joined #openstack-infra | 05:55 | |
| *** dhritishikhar has quit IRC | 05:55 | |
| *** dhritishikhar has joined #openstack-infra | 05:56 | |
| *** sushilkm has joined #openstack-infra | 05:57 | |
| *** sushilkm has left #openstack-infra | 05:57 | |
| *** amitgandhinz has quit IRC | 05:58 | |
| *** otter768 has quit IRC | 05:59 | |
| *** AJaeger has joined #openstack-infra | 06:02 | |
| *** koolhead17 has joined #openstack-infra | 06:06 | |
| *** aswadr has joined #openstack-infra | 06:08 | |
| *** afazekas has joined #openstack-infra | 06:08 | |
| *** mrmartin has quit IRC | 06:12 | |
| *** Guest49005 has quit IRC | 06:13 | |
| *** mrunge has joined #openstack-infra | 06:13 | |
| *** Longgeek has quit IRC | 06:14 | |
| *** Longgeek has joined #openstack-infra | 06:14 | |
| *** yfried has joined #openstack-infra | 06:15 | |
| *** mrunge has quit IRC | 06:16 | |
| *** mrunge has joined #openstack-infra | 06:17 | |
| *** mrunge has quit IRC | 06:17 | |
| AJaeger | ttx, please ping me regarding translations when you're awake and have a few minutes | 06:17 |
| *** sushilkm has joined #openstack-infra | 06:20 | |
| *** sushilkm has left #openstack-infra | 06:21 | |
| *** Ala has joined #openstack-infra | 06:21 | |
| *** yamada-h has joined #openstack-infra | 06:23 | |
| *** koolhead17 has quit IRC | 06:25 | |
| *** jcoufal has joined #openstack-infra | 06:26 | |
| *** yamada-h has quit IRC | 06:27 | |
| openstackgerrit | Jamie Lennox proposed openstack-infra/project-config: Add django-openstack-auth-kerberos project https://review.openstack.org/172803 | 06:29 |
| *** tnovacik has joined #openstack-infra | 06:32 | |
| *** rwsu has joined #openstack-infra | 06:34 | |
| *** scheuran has joined #openstack-infra | 06:37 | |
| *** yfried is now known as yfried|afk | 06:38 | |
| *** yfried|afk is now known as yfried | 06:39 | |
| GheRivero | morning all | 06:40 |
| *** mpaolino has joined #openstack-infra | 06:41 | |
| *** e0ne has joined #openstack-infra | 06:41 | |
| *** ibiris_away is now known as ibiris | 06:41 | |
| *** e0ne has quit IRC | 06:42 | |
| *** yamahata has joined #openstack-infra | 06:42 | |
| *** mrunge has joined #openstack-infra | 06:46 | |
| *** devvesa has joined #openstack-infra | 06:48 | |
| *** yfried is now known as yfried|afk | 06:49 | |
| openstackgerrit | Merged openstack-infra/project-config: Stop running gate-tempest-dsvm-neutron-src-python-keystoneclient-juno https://review.openstack.org/172662 | 06:53 |
| *** soren has joined #openstack-infra | 06:55 | |
| *** yfried|afk is now known as yfried | 06:56 | |
| *** _nadya_ has joined #openstack-infra | 06:57 | |
| *** markus_z has joined #openstack-infra | 06:57 | |
| *** hdd has joined #openstack-infra | 06:58 | |
| *** ajo has joined #openstack-infra | 06:59 | |
| ttx | AJaeger: I'm here | 07:00 |
| AJaeger | morning, ttx! | 07:00 |
| AJaeger | I've run a script over all server projects to check current translation status of files. | 07:00 |
| AJaeger | OUtput is http://paste.openstack.org/show/203412/ | 07:00 |
| ttx | looking | 07:00 |
| AJaeger | The number is just percentages | 07:00 |
| *** hdd has quit IRC | 07:01 | |
| ttx | So the question is... how much is enough for us to keep them ? | 07:02 |
| *** ajo has quit IRC | 07:02 | |
| AJaeger | right now our proposal script imports new translations when they have at least 70 % translated - and removes them again if they have less than 20 %. | 07:03 |
| AJaeger | let me check whether Daisy is around... | 07:03 |
| ttx | Frankly, I'm fine with merging anything the translations team is comfortable with, so I'm happy deferring the % thresholds decisions to you all :) | 07:04 |
| ttx | Although I'd say that keeping a translation that is only 25% just because it once was in... | 07:05 |
| ttx | sounds counterproductive | 07:05 |
| AJaeger | Exactly, I also think it's time for spring cleaning ;) | 07:05 |
| *** Daisy has joined #openstack-infra | 07:05 | |
| ttx | So I'm fine with raising the low threshold | 07:05 |
| Daisy | Hi | 07:05 |
| ttx | 70/50 sounds fair | 07:05 |
| AJaeger | Hi Daisy, thanks for joining! | 07:05 |
| Daisy | :) | 07:05 |
| ttx | At 51 a translation still bring marginal value I suspect | 07:06 |
| ttx | Daisy: Hi! | 07:06 |
| AJaeger | ttx and myself are currently looking at translated files for server projects - and we exclude Horizon here. | 07:06 |
| *** asilenkov_ has quit IRC | 07:06 | |
| *** dteselkin has quit IRC | 07:06 | |
| *** rakhmerov has quit IRC | 07:06 | |
| *** asilenkov has quit IRC | 07:06 | |
| *** HeOS_ has quit IRC | 07:06 | |
| *** katyafervent has quit IRC | 07:06 | |
| *** tsufiev has quit IRC | 07:06 | |
| *** asilenkov has joined #openstack-infra | 07:06 | |
| Daisy | ok. | 07:06 |
| AJaeger | I've run a script over all server projects to check current translation status of files, output is http://paste.openstack.org/show/203412/ | 07:06 |
| openstackgerrit | Jamie Lennox proposed openstack-infra/project-config: Add django-openstack-auth-kerberos project https://review.openstack.org/172803 | 07:06 |
| ttx | Or you keep using 70/20 on the proposal script, but we use.. 66% or something as a release threshold | 07:07 |
| AJaeger | Right now we import new files from transifex once 70 % are translated and propose removing them again once less than 20 % are translated | 07:07 |
| ttx | I don't remember what we used for Juno... 70% ? | 07:07 |
| *** HeOS_ has joined #openstack-infra | 07:07 | |
| *** katyafervent has joined #openstack-infra | 07:07 | |
| AJaeger | Daisy, so, what shall we do for release time now? Change the proposal bot to go with a higher number and let it remove those? Or do a one-time cleanup? | 07:07 |
| *** tsufiev has joined #openstack-infra | 07:07 | |
| AJaeger | ttx, I don't remember, let me check... | 07:08 |
| Daisy | 70% means the completion rate of a single resource ? or a language for a project ? | 07:08 |
| AJaeger | 70 % of single resource | 07:08 |
| ttx | AJaeger: given we now are on proposed/* branches, it's probably more productive to do one-time cleanups | 07:08 |
| openstackgerrit | Jamie Lennox proposed openstack-infra/project-config: Add django-openstack-auth-kerberos project https://review.openstack.org/172803 | 07:09 |
| Daisy | AJaeger: you want to do a one-time cleanups ? | 07:09 |
| ttx | since that lets us merge last-minute updates as well | 07:09 |
| Daisy | I don't have strong opinions to 70% and 20%. I think, they are acceptable. | 07:09 |
| AJaeger | Daisy, I'm wondering whether a 20 % translated resource has any benefit to users, so would raise this for the release | 07:10 |
| Daisy | ok. | 07:10 |
| AJaeger | ttx, Daisy : It's 75 % not 70 % | 07:10 |
| ttx | ok | 07:11 |
| AJaeger | ttx, amotiki is taking care of horizon, so I explictely ignore that one... | 07:11 |
| AJaeger | there they even go for complete translation AFAIR | 07:11 |
| *** jamielennox is now known as jamielennox|away | 07:11 | |
| Daisy | I think 20% translated resources don't have much benefit. | 07:11 |
| *** HeOS has joined #openstack-infra | 07:12 | |
| *** rakhmerov has joined #openstack-infra | 07:12 | |
| Daisy | Even less than 50% translated resources don't have much benefit. That's my opinion. | 07:12 |
| AJaeger | Daisy, agreed. So, which ones should we delete for the release? Less than 50 % ? Less than 75 %? | 07:12 |
| ttx | Yes. For release threshold I'd say 66% is the bare minimum | 07:12 |
| Daisy | :) | 07:12 |
| AJaeger | 66 it is ;) | 07:13 |
| Daisy | There is no specific number. | 07:13 |
| Daisy | I like 66% | 07:13 |
| AJaeger | ttx, so I have to propose patches for the proposed branches only - or also for master? | 07:13 |
| ttx | I think that lats translations that lost a bit of steam (down from 75%) still get in release | 07:13 |
| ttx | lets* | 07:13 |
| ttx | AJaeger: proposed/kilo only | 07:13 |
| AJaeger | ttx, agreed, having something less than 75 % is nice | 07:13 |
| *** dteselkin has joined #openstack-infra | 07:14 | |
| ttx | Let's keep partial translations in master | 07:14 |
| *** ildikov has joined #openstack-infra | 07:14 | |
| AJaeger | ttx, ok, will do directly... | 07:14 |
| ttx | AJaeger: last time, didn't we also catch up with the latest translations within the same script ? | 07:14 |
| ttx | AJaeger: in which case, it would be idea to run the script and propose the translation release cleanup just before the projects RC2s | 07:15 |
| ttx | (i.e. later this week) | 07:15 |
| *** juggler has quit IRC | 07:15 | |
| AJaeger | ttx, let me do one test patch for review and then I'll do it later this week... | 07:15 |
| ttx | If we don't update any string (and just clean up stale stuff) then now is as good as anytime | 07:16 |
| *** dizquierdo has joined #openstack-infra | 07:16 | |
| ttx | AJaeger: do we already have the script uploaded somewhere ? | 07:16 |
| ttx | AJaeger: if not, feel free to propose it to release-tools | 07:16 |
| AJaeger | The one for generating the numbers? | 07:16 |
| AJaeger | Or the one for importing? | 07:16 |
| ttx | The one for importing / cleaning up on proposed branches | 07:17 |
| AJaeger | I don't think it's uploaded - was done manually last time. Let me propose to release-tools... | 07:17 |
| ttx | cool, thx* | 07:17 |
| *** achanda has quit IRC | 07:21 | |
| *** ihrachyshka has joined #openstack-infra | 07:22 | |
| *** achanda has joined #openstack-infra | 07:22 | |
| *** _nadya_ has quit IRC | 07:22 | |
| *** yamada-h has joined #openstack-infra | 07:23 | |
| *** jistr has joined #openstack-infra | 07:24 | |
| *** yamahata has quit IRC | 07:24 | |
| *** chlong has quit IRC | 07:25 | |
| *** sergsh has joined #openstack-infra | 07:27 | |
| *** yamada-h has quit IRC | 07:28 | |
| *** ihrachyshka has quit IRC | 07:28 | |
| *** ihrachyshka has joined #openstack-infra | 07:30 | |
| *** hashar has joined #openstack-infra | 07:30 | |
| AJaeger | one further question: | 07:31 |
| AJaeger | ttx and Daisy, what shall we do with projects that have only the po (Source file) but no translations. Should we update that one as well? | 07:31 |
| *** mrda is now known as mrda-away | 07:32 | |
| AJaeger | I meant pot files | 07:32 |
| Daisy | Which projects ? | 07:32 |
| Daisy | Maybe they don't in Transifex. | 07:32 |
| AJaeger | sahara for example has no translations at all (at least none more than 20 % ) | 07:33 |
| *** arxcruz has joined #openstack-infra | 07:33 | |
| AJaeger | So, should we update the pot file in the repository so that it reflects current status? I think we should... | 07:33 |
| Daisy | I think, pot should be included in the project. It's a part of the source codes. | 07:34 |
| AJaeger | They are included but not updated - so I will update them as part of the import, even for those that have no translations. | 07:34 |
| Daisy | update local pot while importing from Transifex ? | 07:35 |
| AJaeger | That's what we normally do - but if there's nothing to import, we don't update. | 07:35 |
| Daisy | ok. | 07:35 |
| AJaeger | So, my proposed patch will just update the local pot | 07:35 |
| Daisy | ok. | 07:36 |
| Daisy | agree. | 07:36 |
| *** markus_z has quit IRC | 07:37 | |
| *** mpavone has joined #openstack-infra | 07:37 | |
| *** Hal has joined #openstack-infra | 07:39 | |
| *** fhubik has joined #openstack-infra | 07:40 | |
| *** Hal is now known as Guest83327 | 07:40 | |
| *** mpaolino has quit IRC | 07:40 | |
| *** woodster_ has quit IRC | 07:40 | |
| *** markus_z has joined #openstack-infra | 07:41 | |
| *** amitgandhinz has joined #openstack-infra | 07:43 | |
| *** jlanoux has joined #openstack-infra | 07:47 | |
| *** MaxV has joined #openstack-infra | 07:50 | |
| *** btully has quit IRC | 07:54 | |
| *** otter768 has joined #openstack-infra | 07:55 | |
| AJaeger | ttx, Daisy please review https://review.openstack.org/172824 | 07:55 |
| *** jlibosva has joined #openstack-infra | 07:56 | |
| ttx | AJaeger: lgtm | 07:57 |
| Daisy | AJaeger: I'm OK with the patch. Just one concern: there is no auto import script ? You have to import one project by one project manually ? | 07:57 |
| AJaeger | ttx: Commit message clear as well? | 07:57 |
| AJaeger | Daisy, yes, manually for proposed/kilo | 07:57 |
| AJaeger | The autoimport only works on master | 07:57 |
| Daisy | oh. OK. | 07:57 |
| AJaeger | But I have a script for it now ;) | 07:57 |
| ttx | AJaeger: yep | 07:58 |
| Daisy | ok. | 07:58 |
| ttx | Daisy: It's a one time thing | 07:58 |
| Daisy | ok. got it. | 07:58 |
| AJaeger | ttx, so what day shall I do the imports? | 08:00 |
| *** otter768 has quit IRC | 08:01 | |
| amotoki | AJaeger:just /FYI/ I have a script for Horoizon translation import https://github.com/amotoki/horizon-i18n-tools/blob/master/propose-trans.sh I think you are testing a similar script | 08:01 |
| ttx | AJaeger: that will depend based on project. I'd like to do the import once we open a RC2 window for that project | 08:01 |
| AJaeger | ttx, please ping me - and if I do not ack, send an email... | 08:01 |
| ttx | I'd say at first That should happen between Thursday this week and Thursday next week | 08:01 |
| ttx | at first glance* | 08:01 |
| AJaeger | amotoki, yeah... | 08:03 |
| openstackgerrit | Andreas Jaeger proposed openstack-infra/release-tools: Add script to cleanup of translations https://review.openstack.org/172827 | 08:05 |
| AJaeger | ttx, amotoki, here's ^ my script | 08:05 |
| AJaeger | ttx, ok, waiting on Thursday for your first pings ;) | 08:06 |
| *** sergsh has quit IRC | 08:07 | |
| *** e0ne has joined #openstack-infra | 08:08 | |
| *** e0ne has quit IRC | 08:10 | |
| *** _nadya_ has joined #openstack-infra | 08:11 | |
| *** andreykurilin has joined #openstack-infra | 08:11 | |
| *** dtantsur|afk is now known as dtantsur | 08:14 | |
| *** e0ne has joined #openstack-infra | 08:14 | |
| *** pblaho has joined #openstack-infra | 08:14 | |
| tchaypo | jeblair: so I’ve just added all the python-*client repos to gertty. The UI freezes for ~10s when adding a new subscription, but aside from that it’s working fine. It’s still slow to process the queue, but the queue is currently sitting at 115 and gradually changing rather than consistently increasing. | 08:14 |
| openstackgerrit | Andreas Jaeger proposed openstack-infra/release-tools: Add script to cleanup of translations https://review.openstack.org/172827 | 08:14 |
| *** ihrachyshka has quit IRC | 08:15 | |
| *** fhubik is now known as fhubik_afk | 08:16 | |
| *** AJaeger has quit IRC | 08:17 | |
| *** sergsh has joined #openstack-infra | 08:19 | |
| *** e0ne has quit IRC | 08:19 | |
| *** fhubik_afk is now known as fhubik | 08:19 | |
| *** achanda has quit IRC | 08:22 | |
| *** notnownikki has joined #openstack-infra | 08:28 | |
| *** fhubik is now known as fhubik_afk | 08:28 | |
| *** pal has quit IRC | 08:28 | |
| *** pal has joined #openstack-infra | 08:30 | |
| *** AJaeger has joined #openstack-infra | 08:30 | |
| openstackgerrit | Fabien Boucher proposed openstack-infra/infra-specs: Specification proposal about system-config testing using containers https://review.openstack.org/172833 | 08:31 |
| *** derekh has joined #openstack-infra | 08:32 | |
| *** _nadya_ has quit IRC | 08:32 | |
| *** tjones1 has joined #openstack-infra | 08:34 | |
| *** Ala has quit IRC | 08:34 | |
| *** zz_johnthetubagu is now known as johnthetubaguy | 08:36 | |
| *** Ala has joined #openstack-infra | 08:39 | |
| *** fhubik_afk is now known as fhubik | 08:42 | |
| *** Guest83327 has quit IRC | 08:43 | |
| *** Hal has joined #openstack-infra | 08:43 | |
| *** Hal is now known as Guest76030 | 08:44 | |
| *** dizquierdo has quit IRC | 08:46 | |
| *** jlanoux has quit IRC | 08:52 | |
| *** pelix has joined #openstack-infra | 08:53 | |
| *** hichihara has quit IRC | 08:54 | |
| *** mwhudson has quit IRC | 08:54 | |
| *** dguitarbite has quit IRC | 08:56 | |
| *** dhritishikhar has quit IRC | 08:57 | |
| *** ssam2 has joined #openstack-infra | 08:57 | |
| *** jlanoux has joined #openstack-infra | 08:58 | |
| *** mwhudson has joined #openstack-infra | 08:59 | |
| *** yamamoto has quit IRC | 09:01 | |
| *** jamielennox|away is now known as jamielennox | 09:02 | |
| *** dguitarbite has joined #openstack-infra | 09:03 | |
| *** ildikov has quit IRC | 09:04 | |
| *** salv-orlando has joined #openstack-infra | 09:04 | |
| *** jlibosva has quit IRC | 09:05 | |
| *** Guest76030 has quit IRC | 09:07 | |
| *** salv-orlando has quit IRC | 09:07 | |
| *** jlibosva has joined #openstack-infra | 09:07 | |
| *** ihrachyshka has joined #openstack-infra | 09:07 | |
| *** btully has joined #openstack-infra | 09:09 | |
| *** belmoreira has joined #openstack-infra | 09:10 | |
| *** aswadr has quit IRC | 09:14 | |
| *** btully has quit IRC | 09:14 | |
| *** andymaier has joined #openstack-infra | 09:17 | |
| *** SotK_ has joined #openstack-infra | 09:20 | |
| *** doude has quit IRC | 09:21 | |
| *** doude has joined #openstack-infra | 09:22 | |
| *** doude has quit IRC | 09:22 | |
| *** doude has joined #openstack-infra | 09:22 | |
| *** sergsh has quit IRC | 09:23 | |
| *** SotK has quit IRC | 09:24 | |
| *** SotK_ is now known as SotK | 09:25 | |
| *** _nadya_ has joined #openstack-infra | 09:25 | |
| *** jlibosva has quit IRC | 09:32 | |
| *** freerunner has joined #openstack-infra | 09:33 | |
| *** sergsh has joined #openstack-infra | 09:37 | |
| *** Daisy has quit IRC | 09:38 | |
| *** ZZelle has quit IRC | 09:43 | |
| *** ZZelle has joined #openstack-infra | 09:44 | |
| *** fhubik is now known as fhubik_afk | 09:46 | |
| openstackgerrit | Ricardo Carrillo Cruz proposed openstack-infra/system-config: Decouple server class from nodepool service class https://review.openstack.org/172859 | 09:47 |
| *** _1_Nah has joined #openstack-infra | 09:47 | |
| _1_Nah | hello | 09:47 |
| *** fhubik_afk is now known as fhubik | 09:48 | |
| *** _1_Nah has quit IRC | 09:55 | |
| *** e0ne has joined #openstack-infra | 09:55 | |
| *** otter768 has joined #openstack-infra | 09:56 | |
| *** e0ne has quit IRC | 09:56 | |
| *** dizquierdo has joined #openstack-infra | 09:58 | |
| *** markvoelker has joined #openstack-infra | 09:59 | |
| *** otter768 has quit IRC | 10:01 | |
| *** yfried is now known as yfried|afk | 10:02 | |
| openstackgerrit | Boris Pavlovic proposed openstack-infra/project-config: Move Rally to OpenStack https://review.openstack.org/172701 | 10:03 |
| *** nithyag_ has joined #openstack-infra | 10:05 | |
| *** nithyag__ has quit IRC | 10:07 | |
| *** isviridov_away has joined #openstack-infra | 10:09 | |
| *** sdake has joined #openstack-infra | 10:09 | |
| *** pal has quit IRC | 10:10 | |
| *** isviridov_away is now known as isviridov | 10:12 | |
| *** amitpp has joined #openstack-infra | 10:14 | |
| *** markvoelker has quit IRC | 10:16 | |
| *** e0ne has joined #openstack-infra | 10:17 | |
| *** cdent has joined #openstack-infra | 10:18 | |
| *** yfried|afk is now known as yfried | 10:18 | |
| *** markvoelker has joined #openstack-infra | 10:18 | |
| *** fhubik has quit IRC | 10:19 | |
| *** tjones1 has left #openstack-infra | 10:19 | |
| *** fhubik has joined #openstack-infra | 10:20 | |
| *** Ala has quit IRC | 10:21 | |
| *** amitpp has quit IRC | 10:22 | |
| *** isviridov is now known as isviridov_away | 10:23 | |
| *** yamada-h has joined #openstack-infra | 10:25 | |
| *** Ala has joined #openstack-infra | 10:27 | |
| *** marcusvrn has joined #openstack-infra | 10:27 | |
| openstackgerrit | Ricardo Carrillo Cruz proposed openstack-infra/system-config: Decouple openstack_project::server from puppetdb class https://review.openstack.org/172871 | 10:27 |
| *** mrmartin has joined #openstack-infra | 10:27 | |
| *** fhubik_afk has joined #openstack-infra | 10:27 | |
| *** fhubik_afk has joined #openstack-infra | 10:27 | |
| *** fhubik_afk is now known as fhubik_ | 10:27 | |
| *** sdake_ has joined #openstack-infra | 10:28 | |
| *** fhubik has quit IRC | 10:28 | |
| *** markvoelker has quit IRC | 10:29 | |
| *** gillllliard is now known as gilliard | 10:29 | |
| *** yamada-h has quit IRC | 10:30 | |
| *** jlibosva has joined #openstack-infra | 10:30 | |
| *** sdake has quit IRC | 10:31 | |
| openstackgerrit | Ricardo Carrillo Cruz proposed openstack-infra/system-config: Decouple openstack_project::server from puppetdb class https://review.openstack.org/172871 | 10:31 |
| *** e0ne has quit IRC | 10:35 | |
| *** salv-orlando has joined #openstack-infra | 10:35 | |
| *** e0ne has joined #openstack-infra | 10:37 | |
| openstackgerrit | Merged openstack-infra/storyboard-webclient: Support for tags https://review.openstack.org/166847 | 10:38 |
| *** fhubik_ is now known as fhubik | 10:38 | |
| *** dimsum_ has joined #openstack-infra | 10:39 | |
| openstackgerrit | Ricardo Carrillo Cruz proposed openstack-infra/system-config: Decouple server class from nodepool service class https://review.openstack.org/172859 | 10:40 |
| *** e0ne has quit IRC | 10:41 | |
| *** andreykurilin has quit IRC | 10:43 | |
| *** sdake_ has quit IRC | 10:53 | |
| *** teran has joined #openstack-infra | 10:55 | |
| *** mwagner_lap has quit IRC | 10:57 | |
| openstackgerrit | garyk proposed openstack-infra/project-config: VMware: add ability to create and manage branches https://review.openstack.org/172879 | 11:04 |
| *** garyk has joined #openstack-infra | 11:04 | |
| garyk | AJaeger: hi, are you around? | 11:04 |
| garyk | are there cores around? i have a few question about creating branches - https://review.openstack.org/#/c/172879/ | 11:05 |
| openstackgerrit | Marton Kiss proposed openstack-infra/system-config: Initial commit of ask-staging.o.o site https://review.openstack.org/172880 | 11:11 |
| *** yfried is now known as yfried|afk | 11:14 | |
| *** koolhead17 has joined #openstack-infra | 11:14 | |
| *** dhritishikhar has joined #openstack-infra | 11:22 | |
| *** yfried|afk is now known as yfried | 11:22 | |
| *** fhubik is now known as fhubik_afk | 11:25 | |
| *** _nadya_ has quit IRC | 11:26 | |
| *** tiswanso has joined #openstack-infra | 11:31 | |
| openstackgerrit | Fabien Boucher proposed openstack-infra/system-config: Allow passing puppetmaster_server param for jenkins manifests https://review.openstack.org/170487 | 11:32 |
| *** tiswanso has quit IRC | 11:36 | |
| *** yfried is now known as yfried|afk | 11:40 | |
| *** ldnunes has joined #openstack-infra | 11:40 | |
| *** hashar is now known as hasharNap | 11:41 | |
| AJaeger | garyk, now I am... | 11:43 |
| AJaeger | garyk, your patch looks fine... | 11:43 |
| *** yfried|afk is now known as yfried | 11:44 | |
| *** liusheng has quit IRC | 11:44 | |
| kir | Hi all, have a problem with https://review.openstack.org/#/c/157816/ - no verified -1 inside, but when I look at the list of outgoing reviews - I see V=-1 for this patch. What's wrong? | 11:45 |
| *** samueldmq has joined #openstack-infra | 11:45 | |
| *** amotoki has quit IRC | 11:45 | |
| *** weshay has joined #openstack-infra | 11:47 | |
| AJaeger | kir, where do you see V=-1 ? | 11:47 |
| *** spredzy is now known as spredzy|afk | 11:47 | |
| *** ttx has quit IRC | 11:48 | |
| *** ttx has joined #openstack-infra | 11:48 | |
| *** ttx has quit IRC | 11:49 | |
| *** ttx has joined #openstack-infra | 11:49 | |
| *** kir_ has joined #openstack-infra | 11:49 | |
| kir_ | AJaeger: on the list of my outgoing reviews | 11:50 |
| garyk | AJaeger: thanks. Would it be possible to add me to the vmware-nsx-release team (not sure who controls that) | 11:50 |
| openstackgerrit | Matthieu Huin proposed openstack-infra/infra-specs: Allow reviewstats to be used with any issue tracker https://review.openstack.org/172886 | 11:50 |
| *** kir has quit IRC | 11:51 | |
| kir_ | AJaeger: and also here https://review.openstack.org/#/q/owner:kshileev,n,z | 11:52 |
| AJaeger | garyk, you control it yourself, check https://review.openstack.org/#/admin/projects/stackforge/vmware-nsx,access | 11:52 |
| AJaeger | And then click on nsx-release team - that'S you, isn't it? | 11:52 |
| *** dimsum_ has quit IRC | 11:52 | |
| garyk | AJaeger: thanks. | 11:52 |
| *** _nadya_ has joined #openstack-infra | 11:53 | |
| AJaeger | kir_, that looks strange indeed, seems there's a mismatch. I can't help further, let's wait until infra admins like fungi or clarkb show up and can tell you why 157816 has everything green but your overview not | 11:53 |
| *** yamamoto has joined #openstack-infra | 11:55 | |
| *** fhubik_afk is now known as fhubik | 11:57 | |
| *** otter768 has joined #openstack-infra | 11:57 | |
| *** andreykurilin has joined #openstack-infra | 11:58 | |
| *** kir__ has joined #openstack-infra | 11:58 | |
| *** kir_ has quit IRC | 11:58 | |
| kir__ | AJaeger: ok thnks | 11:58 |
| *** andreykurilin has quit IRC | 11:59 | |
| *** sdake has joined #openstack-infra | 12:00 | |
| *** andreykurilin has joined #openstack-infra | 12:00 | |
| *** kgiusti has joined #openstack-infra | 12:00 | |
| *** sdake_ has joined #openstack-infra | 12:01 | |
| openstackgerrit | Filip Blaha proposed openstack-infra/project-config: Pylint job for Murano project https://review.openstack.org/172887 | 12:01 |
| *** mwagner_lap has joined #openstack-infra | 12:02 | |
| *** otter768 has quit IRC | 12:02 | |
| *** sdake has quit IRC | 12:04 | |
| openstackgerrit | Filip Blaha proposed openstack-infra/project-config: Pylint job for Murano project https://review.openstack.org/172887 | 12:05 |
| *** yamamoto has quit IRC | 12:10 | |
| *** Ala has quit IRC | 12:10 | |
| *** Ala has joined #openstack-infra | 12:12 | |
| *** jamielennox is now known as jamielennox|away | 12:12 | |
| *** Daisy has joined #openstack-infra | 12:13 | |
| *** hdd has joined #openstack-infra | 12:15 | |
| *** Hal has joined #openstack-infra | 12:16 | |
| *** maurosr is now known as maurosr_ | 12:16 | |
| *** Hal is now known as Guest11668 | 12:16 | |
| *** dboik has quit IRC | 12:18 | |
| *** EmilienM|afk is now known as EmilienM | 12:19 | |
| *** rfolco has joined #openstack-infra | 12:19 | |
| *** salv-orl_ has joined #openstack-infra | 12:20 | |
| *** zz_jgrimm has quit IRC | 12:22 | |
| *** zz_ja has quit IRC | 12:22 | |
| *** maurosr_ has quit IRC | 12:22 | |
| *** zz_jgrimm has joined #openstack-infra | 12:24 | |
| *** salv-orlando has quit IRC | 12:24 | |
| *** Guest11668 has quit IRC | 12:24 | |
| openstackgerrit | Jaume Devesa proposed openstack-infra/project-config: Use gerrit reference to clone networking-midonet https://review.openstack.org/172897 | 12:25 |
| *** dtantsur is now known as dtantsur|brb | 12:25 | |
| *** zz_ja has joined #openstack-infra | 12:25 | |
| *** sdake_ has quit IRC | 12:25 | |
| *** sdake has joined #openstack-infra | 12:25 | |
| *** maurosr_ has joined #openstack-infra | 12:25 | |
| *** maurosr_ is now known as maurosr | 12:26 | |
| *** dprince has joined #openstack-infra | 12:26 | |
| *** sdake_ has joined #openstack-infra | 12:27 | |
| openstackgerrit | Matthieu Huin proposed openstack-infra/infra-specs: Allow reviewstats to be used with any issue tracker https://review.openstack.org/172886 | 12:28 |
| *** spredzy|afk is now known as spredzy | 12:28 | |
| *** samueldmq_ has joined #openstack-infra | 12:29 | |
| samueldmq | hi, do we still need to keep python26 jobs ? | 12:29 |
| mordred | samueldmq: we're very close to being able to get rid of them | 12:30 |
| samueldmq | mordred, yeah, I was looking at this (https://github.com/openstack-infra/project-config/blob/master/zuul/layout.yaml#L312-L313) | 12:30 |
| samueldmq | mordred, and was wondering if that time already had come :) | 12:31 |
| *** sdake has quit IRC | 12:31 | |
| *** hdd has quit IRC | 12:32 | |
| *** yamamoto has joined #openstack-infra | 12:33 | |
| AJaeger | samueldmq, juno supports python 2.6 - so once it retires, we can remove them | 12:33 |
| mordred | yeah - but - I believe now that juno is cut, we can stop running them on a majority of things | 12:34 |
| *** BharatK has quit IRC | 12:34 | |
| *** fhubik is now known as fhubik_afk | 12:35 | |
| fungi | AJaeger: kir__: i believe it's from citrix xenserver ci which was de-authorized from voting on tempest late last week after it left a -1. it seems that causes its vote to be hidden in the normal review screen but not in the dashboard view | 12:35 |
| BobBall | Is there a nice tool / way that I can remove all -1 votes for the past 2 weeks using the XenServer CI account? | 12:35 |
| *** rlandy has joined #openstack-infra | 12:36 | |
| AJaeger | fungi, so, one way would be to workaround this would be to upload a new change, correct? | 12:37 |
| *** adalbas has joined #openstack-infra | 12:37 | |
| AJaeger | Unless BobBall and you figure out something nicely ;) | 12:37 |
| AJaeger | fungi, thanks! | 12:37 |
| samueldmq | AJaeger, k thanks .. but when J is officially retired ? | 12:38 |
| fungi | BobBall: if it's a new-style account created via openid login, you can manually delete the entries silently from the reviewer table in the webui. it might also be possible to do something similar via the api | 12:38 |
| *** deepakcs has quit IRC | 12:38 | |
| fungi | samueldmq: not yet identified--it'll likely be discussed at the next summit as usual. my guess is ~9 months | 12:38 |
| BobBall | It's been an account for quite some time now (1.5 years?) so I guess it's not a new style... Shame. | 12:38 |
| fungi | could be as few as 5-6 months i guess | 12:38 |
| *** amitgandhinz has quit IRC | 12:39 | |
| samueldmq | fungi - ah ok, thanks | 12:39 |
| *** andreykurilin has quit IRC | 12:39 | |
| fungi | though it's worth noting, now that we're capping and branching stable versions of all the clients and libraries, we could look at dropping py26 testing from kilo/master of all those | 12:40 |
| *** Daisy has quit IRC | 12:40 | |
| *** dboik has joined #openstack-infra | 12:40 | |
| *** _nadya_ has quit IRC | 12:41 | |
| *** kmartin has quit IRC | 12:43 | |
| *** fhubik_afk is now known as fhubik | 12:43 | |
| *** ttx has quit IRC | 12:45 | |
| *** ttx has joined #openstack-infra | 12:45 | |
| *** ttx has quit IRC | 12:46 | |
| *** ttx has joined #openstack-infra | 12:46 | |
| mordred | fungi: ++ | 12:48 |
| dstufft | down with python 2.6 | 12:49 |
| *** eharney has joined #openstack-infra | 12:50 | |
| *** aysyd has joined #openstack-infra | 12:50 | |
| *** jroll has quit IRC | 12:50 | |
| *** jroll has joined #openstack-infra | 12:50 | |
| mordred | dstufft: yeah, tell me about it | 12:51 |
| dstufft | also down with python 2.7, or python 3, either way | 12:51 |
| dstufft | I just want one python again | 12:51 |
| *** bknudson has quit IRC | 12:51 | |
| *** gordc has joined #openstack-infra | 12:51 | |
| mordred | I think that once we can use python3, we should drop support for 2.7 too | 12:51 |
| mordred | becaue there is no need for both at that point | 12:51 |
| samueldmq | mordred, ++ | 12:52 |
| mordred | I mean, other than client libraries | 12:52 |
| dstufft | openstack would probably become the first thing that's really all in on python3 at that point | 12:52 |
| dstufft | which is kinda cool | 12:52 |
| mordred | yah. | 12:52 |
| dstufft | well the first major thing | 12:52 |
| *** zz_dimtruck is now known as dimtruck | 12:52 | |
| samueldmq | fungi, if we want to do that (dropping py26 testing from kilo/master), how to do so? | 12:52 |
| *** markvoelker has joined #openstack-infra | 12:52 | |
| samueldmq | fungi, removing py26 jobs from all projects on zuul/layout.yaml | 12:52 |
| samueldmq | ? | 12:52 |
| mordred | samueldmq: we'll need to do it with a regex match in the jobs section | 12:53 |
| mordred | samueldmq: because we still need to run 2.6 on stable/juno | 12:53 |
| *** dimsum_ has joined #openstack-infra | 12:53 | |
| mordred | actually - this might be a job for one of the plugins, yeah fungi ? | 12:53 |
| samueldmq | mordred, hmm, add a regex at python26-jobs (https://github.com/openstack-infra/project-config/blob/master/zuul/layout.yaml#L314-L324) | 12:54 |
| mordred | samueldmq: http://git.openstack.org/cgit/openstack-infra/project-config/tree/zuul/openstack_functions.py | 12:54 |
| *** markvoelker_ has joined #openstack-infra | 12:54 | |
| mordred | samueldmq: it's too complex for a simple regex I think | 12:54 |
| fungi | mordred: speaking of just using python 3, http://python-future.org/ | 12:54 |
| *** aysyd has quit IRC | 12:55 | |
| AJaeger | there're some stackforge projects that want to have python26 tested on master, like the fuel projects AFAIR, we should not restrict them. We told them they can test basically until Juno end-of-life. | 12:55 |
| AJaeger | But for OpenStack projects I'm not aware of any special requests | 12:55 |
| samueldmq | mordred, k I will take a look (still getting familiar with this proj) :p | 12:55 |
| samueldmq | (too big) | 12:55 |
| *** aysyd has joined #openstack-infra | 12:55 | |
| mordred | fungi: how is that better/worse than six? any idea? | 12:55 |
| *** baoli has joined #openstack-infra | 12:55 | |
| fungi | samueldmq: basically, we already do that for the api servers. they run their python 3.6 jobs today on stable/icehouse and stable/juno | 12:56 |
| samueldmq | fungi, python 3.6? oO | 12:56 |
| fungi | mordred: better in that you write python 3 and this makes it also run on python 2.6/2.7. six tries to introduce support for python back to 2.4 and attempts to provide you with "a way" for your code to run on both 2.x and 3.x | 12:56 |
| AJaeger | fungi, samueldmq indeed, it's only libraries and clients (and yes 3.6->2.6) | 12:56 |
| mordred | fungi: ah - nod | 12:57 |
| *** _nadya_ has joined #openstack-infra | 12:57 | |
| mordred | fungi: I think that's nice | 12:57 |
| fungi | samueldmq: typo. 2.6 | 12:57 |
| *** dustins has joined #openstack-infra | 12:57 | |
| samueldmq | :) | 12:57 |
| mordred | dhellmann: ^^ have you looked at python-future.org? | 12:57 |
| fungi | mordred: their faq also discusses python-future vs six and the pros/cons | 12:57 |
| *** markvoelker has quit IRC | 12:58 | |
| *** dimsum_ has quit IRC | 12:58 | |
| samueldmq | fungi, mordred so basically six code does an if/else to check the python version and then decide the syntax to use ? | 12:59 |
| *** Krinkle|detached is now known as Krinkle | 13:00 | |
| *** jlibosva has quit IRC | 13:00 | |
| *** openstackgerrit has quit IRC | 13:00 | |
| dstufft | samueldmq: sometimes | 13:00 |
| *** miqui has joined #openstack-infra | 13:01 | |
| dstufft | often times really | 13:01 |
| fungi | it's more like six wants you to write in a special python dialect that, with it's help, will run on 2.x and 3.x interpreters. futures seems to be geared more toward letting you write plan python 3.3+ and get it to also run on 2.6/2.7 | 13:01 |
| mordred | fungi: well, I know dhellmann is planning on making a concerted effort towards 3 this cycle | 13:01 |
| fungi | s/plan/plain/ | 13:02 |
| *** jlibosva has joined #openstack-infra | 13:02 | |
| *** dkranz has joined #openstack-infra | 13:02 | |
| dstufft | I haven't used python-future at all, it does some stuff where it installs different things based on python version, which is a no-go for pip D: | 13:02 |
| dstufft | I just use six most of the time | 13:02 |
| *** openstackgerrit has joined #openstack-infra | 13:03 | |
| openstackgerrit | Matthieu Huin proposed openstack-infra/infra-specs: Allow reviewstats to be used with any issue tracker https://review.openstack.org/172886 | 13:03 |
| *** krtaylor has quit IRC | 13:03 | |
| *** hasharNap is now known as hashar | 13:04 | |
| *** doug-fish has left #openstack-infra | 13:04 | |
| mordred | dstufft: oh gross | 13:05 |
| dstufft | mordred: it more or less dynamically builds up a list of packages (in the import sense) to instal based on py2 or py3 | 13:06 |
| *** spzala has joined #openstack-infra | 13:07 | |
| openstackgerrit | Samuel de Medeiros Queiroz proposed openstack-infra/project-config: Fix typo on the description of check-tripleo job https://review.openstack.org/172904 | 13:07 |
| samueldmq | easy review (nit fix) ^ | 13:08 |
| mordred | samueldmq: thanks, looks good - however, we don't normally do patches to only fix nits in infra ... too many other patches to deal with | 13:09 |
| mordred | samueldmq: btw - where in brazil are you? | 13:09 |
| samueldmq | mordred, ah sorry, still not familiar on how things work on infra :) | 13:10 |
| samueldmq | mordred, I am on the Northeast | 13:10 |
| samueldmq | mordred, Campina Grande (City), Paraíba (State) | 13:10 |
| samueldmq | mordred, actually I am part of a team at an university here | 13:11 |
| *** ricky1 has joined #openstack-infra | 13:11 | |
| mordred | samueldmq: cool! I haven't been to the northeast yet, although a good friend of mine is from Recife and tells me I should visit | 13:13 |
| mordred | samueldmq: I usually make it down to Sao Paulo about once a year though | 13:13 |
| samueldmq | mordred, so you still dont know the best part of Brazil :) | 13:14 |
| *** bknudson has joined #openstack-infra | 13:14 | |
| samueldmq | mordred, we have amazing beaches on the northeast | 13:14 |
| *** Hal has joined #openstack-infra | 13:14 | |
| *** isviridov_away is now known as isviridov | 13:14 | |
| *** Hal is now known as Guest96285 | 13:15 | |
| *** erlon has joined #openstack-infra | 13:15 | |
| *** ayoung has joined #openstack-infra | 13:15 | |
| *** isviridov is now known as isviridov_away | 13:16 | |
| *** isviridov_away is now known as isviridov | 13:17 | |
| *** dimtruck is now known as zz_dimtruck | 13:17 | |
| *** radez_g0n3 is now known as radez | 13:18 | |
| dhellmann | mordred: python-future.org looks interesting, I'll dig into that more | 13:18 |
| fungi | it seems to be _on_ pypi at least, though i haven't tried to pip install and play around with it | 13:18 |
| *** isviridov is now known as isviridov_away | 13:19 | |
| *** smccully has joined #openstack-infra | 13:19 | |
| fungi | looks like they only have sdist tarballs for it, no wheels | 13:19 |
| *** koolhead17 has quit IRC | 13:19 | |
| dhellmann | mordred, fungi, dstufft : there's also https://pypi.python.org/pypi/modernize which I need to look at more | 13:19 |
| *** isviridov_away is now known as isviridov | 13:20 | |
| dhellmann | but at this point i'm really thinking more about surrounding tools than I am the application code itself (how do we make devstack do the right thing, for example) | 13:20 |
| ricky1 | folks, why i keep getting "Could not set 'present' on ensure: Could not locate the pip command" on my puppet agent runs, when the puppet-pip is available as a module on the puppetmaster and that module installs pip package | 13:20 |
| fungi | future seems to have something called "futurize" which does similar things to modernize | 13:20 |
| ricky1 | anyone hit this kind of problem recently? | 13:20 |
| *** dhritishikhar has quit IRC | 13:20 | |
| ricky1 | haven't deployed infra servers lately, never had this problem before | 13:20 |
| *** isviridov is now known as isviridov_away | 13:21 | |
| fungi | ricky1: doesn't sound familiar, but we use http://git.openstack.org/cgit/openstack-infra/system-config/tree/install_puppet.sh to install puppet, and it preinstalls pip for us before running puppet at all | 13:22 |
| fungi | ricky1: so we likely wouldn't run into that upstream | 13:22 |
| openstackgerrit | Merged openstack-infra/project-config: Fix typo on the description of check-tripleo job https://review.openstack.org/172904 | 13:24 |
| ricky1 | fungi: so the puppet you install on the nodes are from install_puppet, regardless of being a puppetmaster or other services? | 13:24 |
| ricky1 | following doc from http://ci.openstack.org/puppet.html#id2 seems outdated then | 13:24 |
| fungi | ricky1: yes | 13:24 |
| fungi | ricky1: what specifically is out of date there? | 13:25 |
| ricky1 | that docs says that a node that will join the puppetmaster you have to install puppet with 'sudo apt-get install puppet' | 13:26 |
| ricky1 | but per your statement earlier, it seems you use install_puppet for all ndoes | 13:26 |
| ricky1 | puppetmaster and other services | 13:26 |
| ricky1 | ? | 13:26 |
| ricky1 | I mean the 'Adding a node' section | 13:26 |
| fungi | ricky1: ahh, yeah the "adding a node" section is probably in need of updating. we don't really perform the steps listed there. we use http://git.openstack.org/cgit/openstack-infra/system-config/tree/launch/README when adding new nodes | 13:27 |
| ricky1 | gah | 13:27 |
| ricky1 | o-k | 13:27 |
| ricky1 | thanks fungi | 13:27 |
| fungi | ricky1: what's in that section is an okay general example, but assumes you can work out the missing bits yourself | 13:28 |
| fungi | instead we've just scripted new server deployments entirely | 13:28 |
| fungi | i'll propose an update to that document now | 13:28 |
| *** peristeri has joined #openstack-infra | 13:29 | |
| openstackgerrit | Jan Provaznik proposed openstack-infra/project-config: Set proper group for for core reviews and release https://review.openstack.org/172916 | 13:30 |
| *** amitgandhinz has joined #openstack-infra | 13:30 | |
| *** andreykurilin has joined #openstack-infra | 13:30 | |
| mestery | Good morning infra! Question for you: For all the stackforge/networking-* branches, how do we create stable branches to map to kilo? My google-fu is failing me. :) | 13:31 |
| *** tiswanso has joined #openstack-infra | 13:32 | |
| *** vdo has quit IRC | 13:32 | |
| *** samueldmq_ has quit IRC | 13:32 | |
| *** sks has quit IRC | 13:32 | |
| *** dhritishikhar has joined #openstack-infra | 13:33 | |
| *** woodster_ has joined #openstack-infra | 13:33 | |
| fungi | mestery: the .*-release teams for those projects should be able to go to, for example, https://review.openstack.org/#/admin/projects/stackforge/networking-odl,branches (substitute the correct repo name as needed) and create them there | 13:33 |
| mestery | fungi: Thank you! Looking now. | 13:34 |
| *** baoli has quit IRC | 13:34 | |
| mestery | fungi: So, if I don't see a create buitton there, I must not be in the release team? Fixing that now. | 13:34 |
| *** amitgandhinz has quit IRC | 13:34 | |
| fungi | mestery: either that or the acl for that project doesn't have "create" permission on refs/heads/* for that group | 13:35 |
| mestery | fungi: That must be it! Looking now. | 13:35 |
| *** kgiusti has quit IRC | 13:35 | |
| ricky1 | fungi: sure, i'm automating the deploy an infra from scratch, so i'm biting those bits frequently and pushing changes as I find them | 13:35 |
| openstackgerrit | Marton Kiss proposed openstack-infra/system-config: Add backup scripts to ask.o.o site https://review.openstack.org/169260 | 13:36 |
| ricky1 | fungi: it's ok, i can work on that doc change | 13:36 |
| ricky1 | leave it with me | 13:36 |
| *** baoli has joined #openstack-infra | 13:37 | |
| fungi | ricky1: okay, i'll discard my commit. have at it | 13:37 |
| fungi | and thanks! | 13:37 |
| openstackgerrit | Kyle Mestery proposed openstack-infra/project-config: Add create acls for networking-[odl,ovn] https://review.openstack.org/172919 | 13:37 |
| fungi | mestery: yep, you'll need that | 13:38 |
| mestery | fungi: thanks :) | 13:38 |
| mestery | fungi: IF you have a moment, this one has been sitting for a while too: https://review.openstack.org/#/c/170927/ | 13:39 |
| mestery | fungi: Creates a check job for networking-ovn | 13:39 |
| fungi | mestery: yep, lgtm | 13:40 |
| *** ihrachyshka has quit IRC | 13:42 | |
| *** annegentle has joined #openstack-infra | 13:44 | |
| *** ajo has joined #openstack-infra | 13:45 | |
| *** markvoelker_ has quit IRC | 13:46 | |
| *** mjturek1 has joined #openstack-infra | 13:46 | |
| garyk | fungi: could you please take a look at https://review.openstack.org/172879 - we would like to be able to create a prosed stable branch to be in sync with neutron... | 13:47 |
| *** ihrachyshka has joined #openstack-infra | 13:47 | |
| *** mjturek1 has left #openstack-infra | 13:47 | |
| mestery | fungi: cool, thanks! | 13:48 |
| fungi | garyk: yep, that looks right | 13:48 |
| openstackgerrit | Merged openstack-infra/project-config: Enable a check job for networking-ovn https://review.openstack.org/170927 | 13:48 |
| *** Kiall has quit IRC | 13:50 | |
| mrmartin | er, fungi: I resolved the merge conflict of ask.o.o backup patch, so need to review from scratch again: https://review.openstack.org/#/c/169260/ | 13:51 |
| *** Kiall has joined #openstack-infra | 13:51 | |
| *** mjturek1 has joined #openstack-infra | 13:51 | |
| *** yamamoto has quit IRC | 13:51 | |
| samueldmq | was that possible to create a kind of meta-infra to run infra changes against it and avoiding changes to break the gate, etc | 13:53 |
| samueldmq | setting up infra with the change and doing some tests, as infra does for other OpenStack projects | 13:54 |
| samueldmq | fungi ^ | 13:54 |
| *** zz_dimtruck is now known as dimtruck | 13:54 | |
| *** dtantsur|brb is now known as dtantsur | 13:56 | |
| fungi | samueldmq: we do a poor pass of that by applying all the classes with --noop since they can't all be co-installed at the moment | 13:57 |
| *** loquacities has quit IRC | 13:57 | |
| fungi | samueldmq: there's been some discussion of trying to do real installs of all the classes as a more thorough integration test, but it's not a trivial problem to solve so progress in that direction will be slow | 13:57 |
| *** otter768 has joined #openstack-infra | 13:58 | |
| *** sdake has joined #openstack-infra | 13:59 | |
| samueldmq | fungi, yeah I see, but challenging and nice to have :) | 13:59 |
| *** fhubik is now known as fhubik_afk | 13:59 | |
| *** loquacities has joined #openstack-infra | 13:59 | |
| fungi | samueldmq: well, there's no shortage of other challenging problems to solve here, so it's a matter of prioritizing them | 13:59 |
| samueldmq | fungi, where can I find them ? is there a specs repo ? | 14:01 |
| *** ssam2 has quit IRC | 14:01 | |
| samueldmq | fungi, openstack-infra/infra-specs :) | 14:01 |
| *** ssam2 has joined #openstack-infra | 14:01 | |
| fungi | samueldmq: http://specs.openstack.org/openstack-infra/infra-specs/ and https://review.openstack.org/#/q/status:open+project:openstack-infra/infra-specs,n,z and https://wiki.openstack.org/wiki/Meetings/InfraTeamMeeting#Agenda_for_next_meeting | 14:02 |
| *** superdan is now known as dansmith | 14:02 | |
| *** sdake_ has quit IRC | 14:02 | |
| samueldmq | fungi, thanks | 14:03 |
| *** otter768 has quit IRC | 14:03 | |
| *** shashankhegde has joined #openstack-infra | 14:04 | |
| openstackgerrit | Matthieu Huin proposed openstack-infra/infra-specs: Allow reviewstats to be used with any issue tracker https://review.openstack.org/172886 | 14:04 |
| *** dboik_ has joined #openstack-infra | 14:05 | |
| *** sabeen has joined #openstack-infra | 14:06 | |
| *** dboik has quit IRC | 14:08 | |
| *** lascii is now known as alaski | 14:09 | |
| *** jerryz has joined #openstack-infra | 14:10 | |
| *** dimsum_ has joined #openstack-infra | 14:10 | |
| *** unicell1 has joined #openstack-infra | 14:10 | |
| *** unicell has quit IRC | 14:11 | |
| *** ajmiller has joined #openstack-infra | 14:15 | |
| *** sdake_ has joined #openstack-infra | 14:18 | |
| *** mattfarina has joined #openstack-infra | 14:19 | |
| *** krtaylor has joined #openstack-infra | 14:20 | |
| *** fhubik_afk is now known as fhubik | 14:20 | |
| *** sdake has quit IRC | 14:21 | |
| ayoung | fungi, what is the definition of rc1-ness for gating purposes? Does it just assume that all the projects have that tag, and use the tag upon checkout when running check/gate against it? | 14:23 |
| openstackgerrit | Ricardo Carrillo Cruz proposed openstack-infra/system-config: New nodes should install puppet from install_puppet.sh https://review.openstack.org/172936 | 14:23 |
| fungi | ayoung: nah, it's that there's a proposed/kilo branch created on the project | 14:25 |
| fungi | ayoung: and then further proposed/kilo changes for that project are integration tested against the proposed/kilo or stable/kilo branches of other projects | 14:25 |
| fungi | ayoung: and at that point master of that repo starts working toward liberty | 14:26 |
| ricky1 | fungi: ^ doc change per previous conversation | 14:26 |
| fungi | ricky1: thanks! | 14:26 |
| ricky1 | for some reason i thought that doc was in infra-manual | 14:26 |
| ayoung | fungi, so a RDO needs to do the same thing, just match the branch name on each project? | 14:26 |
| ricky1 | and was giving rounds and rounds | 14:26 |
| ricky1 | should've checked first obvious thing and look in system-config | 14:26 |
| fungi | ricky1: aha, you're rcarrillocruz too | 14:26 |
| ricky1 | yes, i'm at my parents' | 14:26 |
| ricky1 | rcarrillocruz is my bouncer at my home | 14:27 |
| *** yamamoto has joined #openstack-infra | 14:27 | |
| fungi | i need to figure out how to clone myself too. two of me would get so much more work done than one | 14:27 |
| ricky1 | hah | 14:27 |
| openstackgerrit | Merged openstack-infra/project-config: Add create acls for networking-[odl,ovn] https://review.openstack.org/172919 | 14:29 |
| *** tiswanso has quit IRC | 14:29 | |
| *** jistr is now known as jistr|mtg | 14:29 | |
| *** dboik_ has quit IRC | 14:30 | |
| openstackgerrit | Merged openstack-infra/system-config: Add backup scripts to ask.o.o site https://review.openstack.org/169260 | 14:30 |
| *** dboik has joined #openstack-infra | 14:30 | |
| *** achanda has joined #openstack-infra | 14:31 | |
| *** mrmartin has quit IRC | 14:31 | |
| *** wenlock has joined #openstack-infra | 14:32 | |
| *** wenlock1 has joined #openstack-infra | 14:33 | |
| *** dimsum_ is now known as dims | 14:39 | |
| *** achanda has quit IRC | 14:41 | |
| *** yfried has quit IRC | 14:42 | |
| *** tiswanso has joined #openstack-infra | 14:42 | |
| anteaya | I really think there should be a devices-in-public mode that disables _all_ startup/login/ringing sounds, similar to airplane mode | 14:43 |
| *** tiswanso has quit IRC | 14:43 | |
| *** tiswanso has joined #openstack-infra | 14:44 | |
| *** eantyshev has joined #openstack-infra | 14:44 | |
| *** Longgeek has quit IRC | 14:45 | |
| *** _nadya_ has quit IRC | 14:45 | |
| *** Longgeek has joined #openstack-infra | 14:46 | |
| *** markvoelker has joined #openstack-infra | 14:47 | |
| *** sdake has joined #openstack-infra | 14:47 | |
| *** pkarikh has joined #openstack-infra | 14:47 | |
| *** markvoelker_ has joined #openstack-infra | 14:48 | |
| pkarikh | Hi folks! Could anyone take a look at this patch, please? It already has one +2. https://review.openstack.org/#/c/171117/ | 14:49 |
| *** rkukura has quit IRC | 14:50 | |
| *** sdake_ has quit IRC | 14:50 | |
| *** yamamoto has quit IRC | 14:51 | |
| *** cference_ has joined #openstack-infra | 14:51 | |
| *** markvoelker has quit IRC | 14:52 | |
| cference_ | Hi guys, I'm having a server error when I try to update my mailing address on review.openstack.org, anyone have any idea as to why? | 14:52 |
| *** eantyshev has left #openstack-infra | 14:52 | |
| *** markvoelker has joined #openstack-infra | 14:52 | |
| *** markvoelker_ has quit IRC | 14:52 | |
| anteaya | cference_: can you share the error? paste.openstack.org | 14:52 |
| fungi | cference_: also see https://ask.openstack.org/question/56720 | 14:53 |
| cference_ | anteaya, http://paste.openstack.org/show/203594/ | 14:53 |
| openstackgerrit | Merged openstack-infra/project-config: VMware: add ability to create and manage branches https://review.openstack.org/172879 | 14:54 |
| *** fhubik has quit IRC | 14:54 | |
| anteaya | cference_: yup, follow fungi's link | 14:54 |
| anteaya | <-- not a guy | 14:54 |
| *** rwsu has quit IRC | 14:54 | |
| *** rwsu has joined #openstack-infra | 14:54 | |
| *** fhubik has joined #openstack-infra | 14:55 | |
| *** shashankhegde has quit IRC | 14:55 | |
| cference_ | thanks, I'll try that | 14:55 |
| *** fhubik_afk has joined #openstack-infra | 14:55 | |
| *** mriedem has joined #openstack-infra | 14:55 | |
| anteaya | k | 14:55 |
| *** markvoelker has quit IRC | 14:58 | |
| *** amitgandhinz has joined #openstack-infra | 14:59 | |
| openstackgerrit | Timur Nurlygayanov proposed openstack-infra/project-config: Add pep8 check for haos project https://review.openstack.org/172948 | 14:59 |
| *** asselin has joined #openstack-infra | 14:59 | |
| *** mriedem has quit IRC | 14:59 | |
| *** deepakcs has joined #openstack-infra | 15:00 | |
| cference_ | fungi, anteaya, I wasn't using the correct email between the foundation website and gerrit, thanks for the help! | 15:00 |
| anteaya | cference_: ah, glad you found the issue | 15:01 |
| *** hdd has joined #openstack-infra | 15:02 | |
| *** dboik_ has joined #openstack-infra | 15:03 | |
| *** kmartin has joined #openstack-infra | 15:03 | |
| *** mriedem has joined #openstack-infra | 15:03 | |
| *** dboik has quit IRC | 15:07 | |
| *** stevemar has joined #openstack-infra | 15:07 | |
| *** jistr|mtg is now known as jistr | 15:07 | |
| *** jerryz has quit IRC | 15:07 | |
| *** sdake_ has joined #openstack-infra | 15:07 | |
| *** jlibosva has quit IRC | 15:10 | |
| *** hdd has quit IRC | 15:10 | |
| *** trey has quit IRC | 15:11 | |
| openstackgerrit | David Shrewsbury proposed openstack-infra/shade: Add API method create_subnet() https://review.openstack.org/170217 | 15:11 |
| *** garyk has quit IRC | 15:12 | |
| *** annegentle has quit IRC | 15:12 | |
| openstackgerrit | David Shrewsbury proposed openstack-infra/shade: Add API method create_subnet() https://review.openstack.org/170217 | 15:12 |
| *** harlowja_at_home has joined #openstack-infra | 15:12 | |
| *** annegentle has joined #openstack-infra | 15:13 | |
| *** sdake has quit IRC | 15:13 | |
| openstackgerrit | David Shrewsbury proposed openstack-infra/shade: Add API method delete_subnet() https://review.openstack.org/170477 | 15:14 |
| openstackgerrit | David Shrewsbury proposed openstack-infra/shade: Add API method update_subnet() https://review.openstack.org/170516 | 15:14 |
| openstackgerrit | Julia Kreger proposed openstack-infra/shade: Add update_machine method and operator unit test substrate https://review.openstack.org/171165 | 15:14 |
| *** zz_jgrimm is now known as jgrimm | 15:15 | |
| *** trey has joined #openstack-infra | 15:16 | |
| *** dimtruck is now known as zz_dimtruck | 15:17 | |
| openstackgerrit | David Shrewsbury proposed openstack-infra/shade: Add API method create_subnet() https://review.openstack.org/170217 | 15:17 |
| openstackgerrit | David Shrewsbury proposed openstack-infra/shade: Add API method delete_subnet() https://review.openstack.org/170477 | 15:18 |
| *** kgiusti has joined #openstack-infra | 15:18 | |
| *** afazekas has quit IRC | 15:18 | |
| *** koolhead17 has joined #openstack-infra | 15:18 | |
| openstackgerrit | David Shrewsbury proposed openstack-infra/shade: Add API method update_subnet() https://review.openstack.org/170516 | 15:19 |
| openstackgerrit | David Shrewsbury proposed openstack-infra/shade: Add API method update_subnet() https://review.openstack.org/170516 | 15:21 |
| *** koolhead17 has quit IRC | 15:21 | |
| openstackgerrit | David Shrewsbury proposed openstack-infra/shade: Add get_subnet() method https://review.openstack.org/171760 | 15:22 |
| *** dhritishikhar has quit IRC | 15:22 | |
| *** annegentle has quit IRC | 15:23 | |
| *** arxcruz has quit IRC | 15:24 | |
| *** Kiall has quit IRC | 15:26 | |
| *** Kiall has joined #openstack-infra | 15:26 | |
| *** dangers_away is now known as dangers | 15:27 | |
| *** koolhead17 has joined #openstack-infra | 15:27 | |
| *** hashar has quit IRC | 15:29 | |
| *** stevemar has quit IRC | 15:30 | |
| *** dhritishikhar has joined #openstack-infra | 15:30 | |
| *** stevemar has joined #openstack-infra | 15:30 | |
| *** russellb has quit IRC | 15:30 | |
| *** Ala has quit IRC | 15:30 | |
| *** yamahata has joined #openstack-infra | 15:30 | |
| openstackgerrit | Timur Nurlygayanov proposed openstack-infra/project-config: Add pep8 check for haos project https://review.openstack.org/172948 | 15:32 |
| *** dizquierdo has quit IRC | 15:32 | |
| *** mtanino has joined #openstack-infra | 15:32 | |
| *** aysyd has quit IRC | 15:33 | |
| *** russellb has joined #openstack-infra | 15:33 | |
| *** ZZelle_ has joined #openstack-infra | 15:35 | |
| *** soren has quit IRC | 15:35 | |
| *** koolhead_ has joined #openstack-infra | 15:36 | |
| *** koolhead17 has quit IRC | 15:36 | |
| *** mjturek1 has left #openstack-infra | 15:37 | |
| *** dannywilson has joined #openstack-infra | 15:37 | |
| openstackgerrit | Ricardo Carrillo Cruz proposed openstack-infra/system-config: New nodes should install puppet from install_puppet.sh https://review.openstack.org/172936 | 15:37 |
| *** koolhead17 has joined #openstack-infra | 15:37 | |
| *** aysyd has joined #openstack-infra | 15:37 | |
| *** doug-fish has joined #openstack-infra | 15:40 | |
| *** koolhead_ has quit IRC | 15:40 | |
| anteaya | asselin_: thanks | 15:40 |
| anteaya | fungi: all tied up, thanks for being my backup | 15:41 |
| *** scheuran has quit IRC | 15:41 | |
| fungi | anteaya: no worries--glad is worked out! | 15:41 |
| fungi | er, it | 15:41 |
| anteaya | :) | 15:41 |
| *** gyee has joined #openstack-infra | 15:41 | |
| openstackgerrit | Merged openstack-infra/project-config: Update jobs for Merlin https://review.openstack.org/171117 | 15:41 |
| *** spredzy is now known as spredzy|afk | 15:42 | |
| *** rkukura has joined #openstack-infra | 15:42 | |
| *** claudiub has joined #openstack-infra | 15:42 | |
| *** dannywilson has quit IRC | 15:42 | |
| *** mdenny has joined #openstack-infra | 15:43 | |
| *** GheRivero_ has quit IRC | 15:43 | |
| *** dannywilson has joined #openstack-infra | 15:43 | |
| *** garyk has joined #openstack-infra | 15:43 | |
| *** jlibosva has joined #openstack-infra | 15:44 | |
| *** jistr has quit IRC | 15:45 | |
| *** asselin has quit IRC | 15:49 | |
| *** yamahata has quit IRC | 15:49 | |
| *** aricg has left #openstack-infra | 15:51 | |
| *** dannywilson has quit IRC | 15:53 | |
| *** markvoelker_ has joined #openstack-infra | 15:53 | |
| cinerama | hi folks | 15:54 |
| zaro | morning | 15:55 |
| anteaya | hi cinerama | 15:56 |
| anteaya | morning zaro | 15:56 |
| *** ttx has quit IRC | 15:57 | |
| *** ttx has joined #openstack-infra | 15:57 | |
| *** markus_z has quit IRC | 15:57 | |
| *** otter768 has joined #openstack-infra | 15:59 | |
| *** markvoelker_ has quit IRC | 15:59 | |
| mordred | morning cinerama | 16:00 |
| *** annegentle has joined #openstack-infra | 16:00 | |
| anteaya | morning mordred | 16:00 |
| openstackgerrit | Khai Do proposed openstack-infra/puppet-gerrit: Update bouncy castle libs for Gerrit 2.10 https://review.openstack.org/172534 | 16:00 |
| *** dboik_ has quit IRC | 16:01 | |
| cinerama | is there someone who can do Things™ on translate-dev for me? | 16:01 |
| mordred | fungi, clarkb: when you get a chance, eyeballs on 172600 would be helpful - it's getting a bit long and I'd like to land it | 16:01 |
| mordred | anteaya: morning! | 16:01 |
| fungi | mordred: sure thing. i could use a break from wrestling with restructuredtext | 16:02 |
| anteaya | zaro: how goes the standing up of 2.10? is there hope? | 16:02 |
| *** dboik has joined #openstack-infra | 16:02 | |
| anteaya | cinerama: I am unable to do anything on the translate-dev server, sorry | 16:02 |
| zaro | anteaya: looking good. review-dev.o.o is already running 2.10 | 16:03 |
| *** jlibosva has quit IRC | 16:03 | |
| anteaya | zaro: nice work | 16:03 |
| zaro | anteaya: it also has the close connection command | 16:03 |
| mordred | fungi: I'll warn you - there are decorators and function attributes | 16:03 |
| * anteaya goes to check out review-dev | 16:03 | |
| fungi | i'll live | 16:03 |
| *** stevemar has quit IRC | 16:03 | |
| anteaya | zaro: oh yay! | 16:03 |
| mordred | cinerama: what sort of Things do you need? | 16:03 |
| anteaya | \o/ | 16:03 |
| *** otter768 has quit IRC | 16:03 | |
| *** teran_ has joined #openstack-infra | 16:04 | |
| *** stevemar has joined #openstack-infra | 16:04 | |
| *** Swami has joined #openstack-infra | 16:04 | |
| cinerama | mordred: well, first i need to get https://review.openstack.org/#/c/172578/ landed. then applied. and i may need manual mysql intervention | 16:04 |
| *** zz_dimtruck is now known as dimtruck | 16:04 | |
| *** jlanoux has quit IRC | 16:04 | |
| zaro | anteaya: new screen works well in 2.10 if you want to give it a go | 16:05 |
| *** ttx has quit IRC | 16:05 | |
| *** ttx has joined #openstack-infra | 16:05 | |
| anteaya | ohhhh | 16:05 |
| fungi | mordred: ooh, that's quite a deep change stack. i'll see how far i get | 16:05 |
| * anteaya tries new screen | 16:05 | |
| mordred | fungi: I recommend a skim-through read first | 16:05 |
| anteaya | zaro: it appears to default to new screen | 16:06 |
| fungi | indeed | 16:06 |
| mordred | fungi: it's all essnetially smallish changes to get to a sane state - but reading each out of context may be mind bending | 16:06 |
| *** sushilkm has joined #openstack-infra | 16:06 | |
| *** sushilkm has left #openstack-infra | 16:06 | |
| *** fhubik_afk has quit IRC | 16:06 | |
| fungi | starting from the bottom of the stack, we'll see if it causes me to break for lunch instead ;) | 16:06 |
| mordred | cinerama: landed. let's wait for it apply | 16:07 |
| zaro | anteaya: hmm, the preferences menu says default is the old screen | 16:07 |
| *** teran has quit IRC | 16:08 | |
| *** tiswanso_ has joined #openstack-infra | 16:08 | |
| zaro | fungi: ^ would you like to see what you get? | 16:08 |
| anteaya | zaro: really? that does not match my experience | 16:09 |
| zaro | anteaya: look in settings->preferences->change view | 16:09 |
| fungi | anteaya: i wonder if you'd previously adjusted your preference for your account on review-dev when testing the new change screen at some point in the past? | 16:09 |
| openstackgerrit | Clark Boylan proposed openstack-infra/devstack-gate: WIP: Make multinode first class with ansible https://review.openstack.org/172614 | 16:09 |
| *** unicell1 has quit IRC | 16:09 | |
| anteaya | fungi: worth wondering, it is possible, though unlikely as I don't like the new screen, but I can't say for sure | 16:10 |
| anteaya | as I can't remember the last time I was in settings on review-dev | 16:10 |
| fungi | yeah, my preferences are set to "Server Default (Old Screen)" and i'm also seeing the old change screen when browsing around | 16:10 |
| *** afazekas has joined #openstack-infra | 16:10 | |
| anteaya | however if your expierence matches, zaro's, then perhaps it is just me | 16:10 |
| anteaya | okay, just me | 16:10 |
| clarkb | dansmith: https://review.openstack.org/172614 runs the multihost setup in parallel | 16:10 |
| anteaya | thanks for confirming | 16:10 |
| anteaya | hey clarkb | 16:10 |
| clarkb | good morning | 16:11 |
| dansmith | clarkb: ooh nice | 16:11 |
| anteaya | zaro: http://imgur.com/ZTHXR3q | 16:11 |
| anteaya | zaro: so I have a question about athe pull option | 16:11 |
| anteaya | zaro: should we be looking for a way to disable pull? | 16:11 |
| anteaya | or just leave it? | 16:11 |
| *** tiswanso has quit IRC | 16:11 | |
| zaro | anteaya: why do you think we should disable? | 16:13 |
| openstackgerrit | Kyle Mestery proposed openstack-infra/project-config: Add neutron-release as branch creation group https://review.openstack.org/172982 | 16:13 |
| * clarkb points out that 160155 still has unaddressed comments from ps1 | 16:13 | |
| clarkb | mordred: ^ | 16:13 |
| anteaya | zaro: well pull isn't really in our workflow, will that mess up some devs if they choose to pull? | 16:14 |
| * anteaya thinks backto draft | 16:14 | |
| zaro | anteaya: i do pull all the time :) | 16:15 |
| clarkb | the potential problem with pull is it does an implicit merge by default | 16:15 |
| mordred | clarkb: yah - I've been ignoring that one until it fixes its merge conflict state | 16:15 |
| clarkb | mordred: ya I am just being cranky that we are trying to push through a giant stack when an ancient change has been ignored instead | 16:15 |
| openstackgerrit | Merged openstack-infra/system-config: Add default from address https://review.openstack.org/172578 | 16:16 |
| mordred | anteaya, zaro; I REALLY wish it was possible to configure pull to use --ff-only by default | 16:16 |
| clarkb | (all of which add tests) | 16:16 |
| anteaya | clarkb: yes, thanks that was the issue I was thinking of but couldn't articulate | 16:16 |
| mordred | clarkb: would it help your cranky if I abandoned that change? | 16:16 |
| openstackgerrit | Matthew Treinish proposed openstack/os-testr: Add support for having comments in the exclude file https://review.openstack.org/172984 | 16:16 |
| mordred | because we've already added create_server unittests in a different patch | 16:17 |
| anteaya | zaro: is it possible to configure pull to use --ff-only by default? | 16:17 |
| *** smccully has quit IRC | 16:17 | |
| clarkb | mordred: if it has been superceded then yes we should likely abandon it | 16:17 |
| mrodden | i think you can get it to --rebase by default... | 16:17 |
| *** smccully has joined #openstack-infra | 16:17 | |
| mordred | clarkb: done | 16:17 |
| anteaya | mordred: would pull --rebase by default be acceptable for pull? | 16:18 |
| clarkb | the pull implicit merge is by the way one of the reasons why I think people think git review does a rebase autoamtically even when not necessary | 16:18 |
| *** Longgeek has quit IRC | 16:18 | |
| anteaya | clarkb: oh | 16:18 |
| clarkb | anteaya: mordred ^ and because of that I don't think any side effecting behavior of pull is desirable | 16:18 |
| *** belmoreira has quit IRC | 16:18 | |
| clarkb | rebase vs ff only doesn't matter, it still causes confusion and people yell at us to fix bugs that don't exist | 16:19 |
| mordred | clarkb: yah. agree | 16:19 |
| anteaya | clarkb: what would you like to see? right now we have: http://imgur.com/ZTHXR3q | 16:19 |
| mordred | clarkb: I didn't mean for our docs | 16:19 |
| mordred | I mean that I personally wish I could set that locally | 16:19 |
| fungi | we should probably make checkout the default option anyway | 16:19 |
| *** tqtran has joined #openstack-infra | 16:19 | |
| *** ayoung is now known as dreadadminrobert | 16:19 | |
| mordred | ++ | 16:19 |
| *** dreadadminrobert is now known as ayoung | 16:19 | |
| *** dhritishikhar has quit IRC | 16:19 | |
| *** tiswanso_ has quit IRC | 16:19 | |
| clarkb | anteaya: I am a fan of checkout, it is what I use locally | 16:19 |
| mordred | the two things I do the most are checkout and reset --hard | 16:20 |
| anteaya | me too | 16:20 |
| *** btully has joined #openstack-infra | 16:20 | |
| anteaya | the think I'm think is any lever we expose gets used | 16:20 |
| *** carl_baldwin_ has joined #openstack-infra | 16:20 | |
| *** tiswanso has joined #openstack-infra | 16:20 | |
| mordred | yup - which means if we can turn pull off, that's awesome | 16:20 |
| clarkb | yes the other git review trouble maker is git review -x | 16:20 |
| anteaya | so should we try to unexpose the levers we don't want to fix afterthey are used? | 16:20 |
| mordred | I use git review -x CONSTANTLY | 16:20 |
| *** ricky1 has quit IRC | 16:20 | |
| clarkb | mordred: yes but you understand git | 16:20 |
| mordred | clarkb: yup | 16:20 |
| clarkb | mordred: most people odn't understand that you get a new sha1 every time you use it | 16:21 |
| mordred | sigh | 16:21 |
| mordred | well, I could live without it | 16:21 |
| clarkb | mordred: this results in two problems. 1) gerrit rejects pushes because there are no changes | 16:21 |
| mtreinish | clarkb: wait, really? sigh... | 16:21 |
| clarkb | 2) it looks like git review did an implicit rebase for you when reall you just cherry picked rewriting all history yourself | 16:21 |
| mordred | maybe we should make an expert edition of git review | 16:21 |
| *** Somay has joined #openstack-infra | 16:21 | |
| anteaya | oh great, more levels | 16:22 |
| *** dboik has quit IRC | 16:22 | |
| mordred | because clarkb makes a good point - it confuses people ... otoh, it's super helpful if you know what you're doing | 16:22 |
| zaro | you mean just remove pull from UI right? | 16:22 |
| mordred | zaro: yah | 16:22 |
| clarkb | mordred: also fwiw I never use git review -x | 16:23 |
| clarkb | its usually just as easy/hard to git review -d then rebase | 16:23 |
| mordred | clarkb: sometimes I'm not rebasing | 16:23 |
| zaro | if that's not an option in Gerrit core, then you can probably do the javascript thing after page render | 16:23 |
| mtreinish | I use it all the time, too | 16:23 |
| clarkb | but ^ comes with the added benefit of git holding my hand | 16:23 |
| mordred | clarkb: this comes back to the thing that we still don't have good support for | 16:23 |
| mordred | clarkb: which is constructing a local repo state from several unrealted upstream changesets | 16:24 |
| clarkb | mordred: sure, and rebase can do that for you | 16:24 |
| *** Sukhdev has joined #openstack-infra | 16:24 | |
| mordred | clarkb: I actually gave up and started using merge in a throwaway branch | 16:24 |
| fungi | i use git review -x when i'm adding various commits to add dependency relationships which did not previously exist. i could use git review -d instead and then rebase -i and add pick lines | 16:24 |
| * anteaya thinks she needs to see clarkb's tutorial | 16:24 | |
| clarkb | the main reason I use rebase is due to conflict handling | 16:24 |
| clarkb | dealing with conflicts when not in a merge/rebase is weird | 16:24 |
| mordred | clarkb: but the dancing I have to do currently to deal with and test the dib-nodepool stuff is BANANAS | 16:24 |
| *** dannywilson has joined #openstack-infra | 16:25 | |
| *** dannywilson has quit IRC | 16:25 | |
| * anteaya would like to see mordred dancing | 16:25 | |
| clarkb | fungi: right you just fetch the refs then rebase onto master and pick sha1s :) | 16:25 |
| mordred | becasue there are multiple outstanding changes across multiple repos that all need to be present for anything to work | 16:25 |
| zaro | here you can customize that download mneu: https://gerrit.googlesource.com/plugins/download-commands/+doc/master/src/main/resources/Documentation/config.md | 16:25 |
| fungi | the version of git i use makes merge conflict resolution effectively the same for cherry-pick as for merge or rebase. i don't really see a difference where that's concerned | 16:25 |
| clarkb | fungi: which is effectivel the same work as git review -x sha1 | 16:25 |
| *** dannywilson has joined #openstack-infra | 16:25 | |
| clarkb | fungi: a cherry pick is an all or nothing state, its not a stack like a rebease | 16:26 |
| krotscheck | Can some infra cores go through the storyboard specifications and start weighing in? Some of these date from January. | 16:26 |
| clarkb | the end resutls are the same but rebase has a few more tools | 16:26 |
| *** hdd has joined #openstack-infra | 16:26 | |
| fungi | there's an incomplete cherry-pick state. you fix and add your conflict resolution, then --continue | 16:26 |
| fungi | seems basically the same to me | 16:26 |
| *** vhoward has left #openstack-infra | 16:26 | |
| anteaya | zaro: nice, so removing pull is easy | 16:26 |
| mordred | fungi: git add /git commit - not continue - but yes - I also do not find it confusing | 16:26 |
| clarkb | fungi: hrm is that new? I don't think git that was in precise had that | 16:26 |
| *** dboik has joined #openstack-infra | 16:26 | |
| anteaya | zaro: what is format patch? | 16:27 |
| zaro | yep. just a gerrit.config setting | 16:27 |
| clarkb | fungi: it was either commit or reset --hard | 16:27 |
| harlowja_at_home | could i get some folks to look @ https://review.openstack.org/#/c/171393/ + dependents that add tests (a release would be cool to) so that i can merge some other code that needs to be able to shutdown that thread for testing.... | 16:27 |
| fungi | mordred: huh... i use git cherry-pick --continue all the time | 16:27 |
| mordred | fungi: REALLY? neat | 16:27 |
| zaro | anteaya: no idea, never used that command | 16:28 |
| openstackgerrit | Merged openstack-infra/shade: Add tests for caching https://review.openstack.org/171798 | 16:28 |
| clarkb | hrm I must be thinking of stash not cherry pick | 16:28 |
| anteaya | zaro: me ethier, any objection to removing that too? we can add it bak if we need it | 16:28 |
| mordred | zaro, anteaya: I believe we also do not want format-patch | 16:28 |
| mordred | it's somethign people use for email-based workflows | 16:28 |
| anteaya | mordred: oh good | 16:29 |
| zaro | anteaya: i think this is it https://git-scm.herokuapp.com/docs/git-format-patch | 16:29 |
| anteaya | ah | 16:29 |
| mordred | we also probably don't want "repo download" | 16:29 |
| *** MaxV has quit IRC | 16:29 | |
| fungi | clarkb: yeah, stash/stash pop is definitely a little stranger | 16:29 |
| mordred | conflicts on stash pop are fun | 16:29 |
| * anteaya is being forced by the train people to eat lunch now | 16:29 | |
| *** afazekas has quit IRC | 16:30 | |
| *** tiswanso has quit IRC | 16:30 | |
| zaro | anteaya: i don't see 'repo download' | 16:31 |
| *** baoli has quit IRC | 16:31 | |
| *** tiswanso_ has joined #openstack-infra | 16:32 | |
| andreaf | gerrit question: if I hit +A by mistake in gerrit, and it triggers gate, will it still merge even if I removed the +A? Or do I have to -A or -2? | 16:32 |
| mordred | zaro: | 16:32 |
| andreaf | https://review.openstack.org/#/c/146535 ^^^ | 16:32 |
| mordred | scheme = repo_download | 16:32 |
| clarkb | andreaf: it will merge if tests pass and it has a +2 code review +1 workflow and no -2s | 16:32 |
| clarkb | andreaf: so if you remove your +1 workflow and there are no other +1 workflows it will not merge | 16:32 |
| andreaf | ok, so dropping the +1 on workflow is enough | 16:33 |
| andreaf | clarkb: thanks | 16:33 |
| *** baoli_ has joined #openstack-infra | 16:33 | |
| *** pblaho has quit IRC | 16:33 | |
| *** sandywalsh has quit IRC | 16:34 | |
| zaro | andreaf: i see, that's already off by default | 16:34 |
| *** dougwig has quit IRC | 16:35 | |
| *** hogepodge has quit IRC | 16:35 | |
| *** sandywalsh has joined #openstack-infra | 16:35 | |
| *** dougwig has joined #openstack-infra | 16:35 | |
| *** sergsh has quit IRC | 16:36 | |
| *** patrickeast has joined #openstack-infra | 16:37 | |
| *** MaxV has joined #openstack-infra | 16:37 | |
| clarkb | dansmith: btw live migration tests seem to fail in a tempest check. mtreinish looked at it briefly on friday and saw something in the nova compute log that he thought might be related | 16:39 |
| *** aysyd has quit IRC | 16:39 | |
| dansmith | clarkb: okay, all of them or just some of them? | 16:39 |
| clarkb | dansmith: it seems to be consistent | 16:40 |
| clarkb | dansmith: let me get you a link to the logs | 16:40 |
| dansmith | okay | 16:40 |
| *** sputnik13 has joined #openstack-infra | 16:40 | |
| clarkb | dansmith: http://logs.openstack.org/14/172614/13/experimental/check-tempest-dsvm-aiopcpu-full/599095f/console.html#_2015-04-11_19_32_43_339 | 16:40 |
| *** bknudson has quit IRC | 16:40 | |
| *** dhritishikhar has joined #openstack-infra | 16:41 | |
| *** subscope_ has joined #openstack-infra | 16:41 | |
| dansmith | clarkb: ah, yeah, the console log failure | 16:42 |
| dansmith | hmm | 16:42 |
| *** dannywilson has quit IRC | 16:42 | |
| *** aysyd has joined #openstack-infra | 16:42 | |
| *** koolhead17 has quit IRC | 16:43 | |
| *** unicell has joined #openstack-infra | 16:43 | |
| *** Guest96285 has quit IRC | 16:43 | |
| *** mpavone has quit IRC | 16:44 | |
| *** devvesa has quit IRC | 16:45 | |
| zaro | anybody know anything about mime-type? like why would Gerrit recognize 'text/x-yaml' but not 'text/yaml' ? | 16:46 |
| *** unicell has quit IRC | 16:46 | |
| *** unicell has joined #openstack-infra | 16:46 | |
| *** harlowja_at_home has quit IRC | 16:46 | |
| anteaya | zaro: I do not know the answer to your question | 16:46 |
| *** kgiusti has quit IRC | 16:47 | |
| openstackgerrit | Amaury Medeiros proposed openstack-infra/release-tools: Use fill method to simplify string wrapping https://review.openstack.org/172993 | 16:47 |
| sdague | zaro: text/yaml might not have an IANNL registration - http://www.iana.org/assignments/media-types/media-types.xhtml | 16:47 |
| anteaya | zaro: the 'Open All' button appears to open all the separate files in the patch in new tabs, is this the expected behaviour? | 16:48 |
| *** ricky1 has joined #openstack-infra | 16:49 | |
| zaro | anteaya: yeah, it's a bad featuer | 16:49 |
| anteaya | agreed, as long is it isn't a bug | 16:50 |
| zaro | it should be open all files in one page -> single page review | 16:50 |
| anteaya | yes, that would be a good feature | 16:50 |
| zaro | sdague: thanks | 16:50 |
| *** shashankhegde has joined #openstack-infra | 16:50 | |
| mordred | zaro: someone has been working on that recently, yeah? | 16:50 |
| mordred | zaro: I thought I saw you reviewing their patches at one point | 16:50 |
| openstackgerrit | Merged openstack-infra/shade: Refactor caching to allow per-method invalidate https://review.openstack.org/171456 | 16:50 |
| *** ivar-lazzaro has joined #openstack-infra | 16:51 | |
| anteaya | 'Toggle CI' button is working as expected for me | 16:51 |
| zaro | mordred: yeah, it was me but it died because old screen got ripped out. | 16:51 |
| mordred | sigh | 16:52 |
| mordred | wow, I'm SO GLAD effort was put towards old screen and not towards single-page-diff | 16:52 |
| anteaya | in expanding comments, I see one button 'Expand All', which becomes 'Collapse All' when expanded | 16:52 |
| *** ricky1 has quit IRC | 16:52 | |
| mordred | s/old/new/ | 16:52 |
| anteaya | I'm not seeing a 'Expand Recent' button or feature | 16:52 |
| *** afazekas has joined #openstack-infra | 16:53 | |
| reed | can I please get a review of this request for a mailing list for coordination of internships? https://review.openstack.org/#/c/167458/ | 16:54 |
| *** tonytan4ever has joined #openstack-infra | 16:55 | |
| * anteaya wonders if this should be one of the things she cares about or doesn't care about anymore | 16:55 | |
| anteaya | referencing 167458 | 16:56 |
| *** vhoward has joined #openstack-infra | 16:57 | |
| reed | anteaya, it's mainly for GSoC and Outreachy at this moment, for mentors and administrators to coordinate | 16:57 |
| *** derekh has quit IRC | 16:57 | |
| *** yamahata has joined #openstack-infra | 16:57 | |
| *** dtantsur is now known as dtantsur|afk | 16:57 | |
| anteaya | yeah I see that | 16:58 |
| *** dimtruck is now known as zz_dimtruck | 17:00 | |
| *** armax has joined #openstack-infra | 17:00 | |
| *** HeOS has quit IRC | 17:00 | |
| openstackgerrit | Merged openstack-infra/storyboard: Add tools to run integration tests https://review.openstack.org/155975 | 17:00 |
| *** ihrachyshka has quit IRC | 17:00 | |
| *** dustins has quit IRC | 17:01 | |
| *** dkranz has quit IRC | 17:01 | |
| *** dustins has joined #openstack-infra | 17:02 | |
| *** Somay has quit IRC | 17:02 | |
| *** marun has joined #openstack-infra | 17:03 | |
| clarkb | speaking of internships, I need to finish prep for my capstone thing | 17:03 |
| *** BharatK has joined #openstack-infra | 17:03 | |
| clarkb | greghaynes: ^ maybe work on that tomorrow? | 17:03 |
| *** flaper87 has quit IRC | 17:03 | |
| BharatK | Hi, Please suggest me the way to skip tempest test cases through job definition | 17:04 |
| BharatK | I want to skip "test_volume_boot_pattern" test cases | 17:04 |
| anteaya | BharatK: I suggest you have a chat with cinder folks about skipping tests | 17:05 |
| anteaya | BharatK: they may have some opinions about it | 17:05 |
| anteaya | and if they do, it is a good idea if you know what they are | 17:05 |
| *** andymaier has quit IRC | 17:06 | |
| BharatK | anteaya: Actuall I sent a patch https://review.openstack.org/#/c/172841/ | 17:06 |
| anteaya | BharatK: well #openstack-qa is the channel tempest developers are present in | 17:06 |
| *** marun has quit IRC | 17:06 | |
| BharatK | anteaya: Okay I will ask there thanks. | 17:06 |
| anteaya | BharatK: if you want to discuss your above patch, that is a good place to chat | 17:06 |
| anteaya | welcome | 17:07 |
| *** imcsk8 is now known as imcsk8|afk | 17:07 | |
| *** subscope_ has quit IRC | 17:08 | |
| *** sputnik13 has quit IRC | 17:08 | |
| *** AJaeger_ has joined #openstack-infra | 17:08 | |
| *** dhritishikhar_ has joined #openstack-infra | 17:08 | |
| *** afazekas has quit IRC | 17:09 | |
| *** dhritishikhar has quit IRC | 17:09 | |
| zaro | mordred: hadn't realized that gerrit doesn't support syntax highlighting for puppet :( | 17:09 |
| *** annegentle has quit IRC | 17:10 | |
| SpamapS | bummer | 17:10 |
| openstackgerrit | Merged openstack-infra/shade: Allow passing config into shade.openstack_cloud https://review.openstack.org/171857 | 17:10 |
| SpamapS | syntax highlighting is like A/C for me. Can I live without it? Yes. Will I be happy? NO. | 17:11 |
| greghaynes | clarkb: ++ | 17:11 |
| SpamapS | (Air conditioning, not alternating current, though same words apply) | 17:11 |
| yolanda | mordred, do you have a sec? i have a pair of points to raise with you | 17:12 |
| *** BobBall is now known as BobBall_AWOL | 17:12 | |
| yolanda | 1. my mentee will start working likely on https://review.openstack.org/155305 , is that fine? | 17:12 |
| yolanda | not a spec with +A , but with general approval | 17:12 |
| *** ssam2 has quit IRC | 17:13 | |
| yolanda | 2. that needs some attention: https://review.openstack.org/170788 | 17:13 |
| *** ibiris is now known as ibiris_away | 17:13 | |
| med_ | "mentee" like it. | 17:13 |
| med_ | never seen it used before, albeit real word. | 17:14 |
| med_ | (yes, yolanda has a better command of English than I do.) | 17:14 |
| yolanda | well, i used google translate for that | 17:14 |
| yolanda | :) | 17:14 |
| yolanda | nice to see you around, med, how are things? :) | 17:16 |
| med_ | very well, thanks yolanda. And you? | 17:16 |
| *** flaper87 has joined #openstack-infra | 17:16 | |
| *** tonytan4ever has quit IRC | 17:16 | |
| yolanda | nice, quite busy all time, but i love the things i'm involved on | 17:17 |
| med_ | xlnt. | 17:17 |
| *** fesp has joined #openstack-infra | 17:18 | |
| mordred | yolanda: looking | 17:18 |
| *** achanda has joined #openstack-infra | 17:18 | |
| *** MaxV has quit IRC | 17:18 | |
| *** davideagnello has joined #openstack-infra | 17:18 | |
| *** MaxV has joined #openstack-infra | 17:19 | |
| mordred | yolanda: yes - that is a good thing- although keep in mind it might be something that wants to be added to shade and exposed in a sane way to nodepool | 17:19 |
| mordred | yolanda: or it might be pure nodepool | 17:19 |
| mordred | *waves hands* | 17:19 |
| mordred | hi med_ ! | 17:19 |
| yolanda | mordred, well, it should be moved to shade, yes | 17:19 |
| yolanda | ah, code changes faster than specs | 17:19 |
| yolanda | so i shall rewrite some bits of that before she starts putting hands on it | 17:20 |
| mordred | yolanda: maybe a good idea? | 17:20 |
| * med_ nods at mordred whom he barely knows | 17:20 | |
| yolanda | yes, it is :) | 17:20 |
| *** fesp has quit IRC | 17:21 | |
| mordred | yolanda: for the second one - I think jeblair wanted to wait on landing that for a bit and see how it works for you locally in gozer land ... but we should wait for him to get back online | 17:21 |
| *** deepakcs_ has joined #openstack-infra | 17:21 | |
| yolanda | ah man, mordred, so you want that we land this in Gozer prior to upstream? :) | 17:21 |
| yolanda | i kept that running on my nodepool account for 2 days, but on a testing account | 17:22 |
| *** deepakcs has quit IRC | 17:22 | |
| *** deepakcs_ is now known as deepakcs | 17:22 | |
| *** dkranz has joined #openstack-infra | 17:23 | |
| yolanda | when is jeblair coming back? | 17:23 |
| *** dhritishikhar_ has quit IRC | 17:23 | |
| *** MaxV has quit IRC | 17:23 | |
| *** hogepodge has joined #openstack-infra | 17:23 | |
| anteaya | yolanda: when he lands I expect | 17:23 |
| *** deepakcs has quit IRC | 17:24 | |
| *** tonytan4ever has joined #openstack-infra | 17:24 | |
| yolanda | i'd like to have his opinion prior to testing in Gozer really, makes sense to wait a bit | 17:25 |
| openstackgerrit | Stuart McLaren proposed openstack-infra/project-config: Stop running gate-tempest-dsvm-neutron-src-python-glanceclient-juno https://review.openstack.org/172999 | 17:25 |
| *** adalbas has quit IRC | 17:25 | |
| anteaya | yolanda: I wouldn't be surprised to see him in a few hours, if not then probably tomorrow | 17:26 |
| yolanda | i won't be online for much longer today, but i'd lke to catch up with him tomorrow | 17:26 |
| *** jcoufal has quit IRC | 17:26 | |
| openstackgerrit | Stuart McLaren proposed openstack-infra/project-config: Stop running gate-tempest-dsvm-neutron-src-python-glanceclient-juno https://review.openstack.org/172999 | 17:27 |
| *** crainte has joined #openstack-infra | 17:27 | |
| openstackgerrit | Merged openstack-infra/shade: Test volume list caching https://review.openstack.org/172586 | 17:27 |
| openstackgerrit | Clark Boylan proposed openstack-infra/devstack-gate: WIP: Make multinode first class with ansible https://review.openstack.org/172614 | 17:27 |
| *** dizquierdo has joined #openstack-infra | 17:27 | |
| fungi | SpamapS: did you have a rebuttal for Shrews's concern on https://review.openstack.org/172587 ? | 17:27 |
| Shrews | fungi: it's all good. i understand it now | 17:28 |
| fungi | k | 17:28 |
| *** SumitNaiksatam has joined #openstack-infra | 17:28 | |
| anteaya | yolanda: yeah, I don't see that waiting will get you much, better trying tomorrow | 17:28 |
| fungi | Shrews: aha, you had already re-commented. i'm slow ;) | 17:29 |
| kragniz | some reviews on https://review.openstack.org/172999 would be nice | 17:31 |
| kragniz | glanceclient gate currently borked | 17:31 |
| *** baoli_ has quit IRC | 17:32 | |
| *** afazekas has joined #openstack-infra | 17:34 | |
| SpamapS | fungi: it's not that you're slow, it's just that we're _so damn fast_ ;) | 17:36 |
| *** johnthetubaguy is now known as zz_johnthetubagu | 17:36 | |
| *** tiswanso_ has quit IRC | 17:36 | |
| fungi | yes. yes you are | 17:37 |
| *** harlowja_away is now known as harlowja | 17:37 | |
| *** baoli_ has joined #openstack-infra | 17:37 | |
| *** tiswanso_ has joined #openstack-infra | 17:37 | |
| bswartz | hello infra! | 17:40 |
| bswartz | I just realized that there is no manila-stable-maint team in gerrit | 17:40 |
| fungi | bswartz: does it need one? do you (expect to) have separate people doing stable maint for manila? | 17:41 |
| bswartz | who should I be talking to about setting one up? | 17:41 |
| bswartz | fungi: well no | 17:41 |
| bswartz | ideally the stable branch would be managed by the same core team | 17:41 |
| *** dboik has quit IRC | 17:41 | |
| bswartz | with help from the stable-maint core team | 17:42 |
| fungi | bswartz: that's what it looks like your acl allows currently | 17:42 |
| bswartz | oh, excellent | 17:42 |
| bswartz | In the past I've had problems +2ing stuff to the stable/juno branch | 17:42 |
| *** dboik has joined #openstack-infra | 17:42 | |
| fungi | bswartz: have you tried in the past couple months since we decentralized control over stable branches? | 17:42 |
| anteaya | http://git.openstack.org/cgit/openstack-infra/project-config/tree/gerrit/acls/openstack/manila.config | 17:42 |
| bswartz | fungi: no | 17:42 |
| fungi | i think you should be fine now, but if not let me know what's missing | 17:43 |
| bswartz | okay this looks like an improvement then since last I checked | 17:43 |
| bswartz | thanks | 17:43 |
| fungi | yep. moar tent | 17:43 |
| fungi | s/tent/yurt/ | 17:43 |
| * fungi keeps forgetting | 17:43 | |
| *** tonytan_brb has joined #openstack-infra | 17:43 | |
| anteaya | is it yurt now? | 17:43 |
| * anteaya updates | 17:44 | |
| bswartz | do yurts scale better than tents? | 17:44 |
| fungi | everywhere except at the circus | 17:45 |
| *** tonytan4ever has quit IRC | 17:45 | |
| *** marun has joined #openstack-infra | 17:45 | |
| *** teran_ has quit IRC | 17:45 | |
| *** teran has joined #openstack-infra | 17:45 | |
| *** dhritishikhar_ has joined #openstack-infra | 17:47 | |
| anteaya | the hard part with the yurt is building the stabilizing circle for the roof | 17:47 |
| anteaya | the wall are easy | 17:47 |
| anteaya | walls | 17:47 |
| *** bknudson has joined #openstack-infra | 17:48 | |
| *** BharatK has quit IRC | 17:49 | |
| *** r-daneel has joined #openstack-infra | 17:51 | |
| *** aysyd has quit IRC | 17:51 | |
| openstackgerrit | Merged openstack-infra/shade: Unsteady state in volume list should prevent cache https://review.openstack.org/172587 | 17:52 |
| *** dizquierdo has quit IRC | 17:52 | |
| *** zz_dimtruck is now known as dimtruck | 17:52 | |
| *** dizquierdo has joined #openstack-infra | 17:52 | |
| *** aysyd has joined #openstack-infra | 17:53 | |
| openstackgerrit | Clint 'SpamapS' Byrum proposed openstack-infra/shade: Add test for invalidation after delete https://review.openstack.org/172600 | 17:54 |
| *** dboik_ has joined #openstack-infra | 17:55 | |
| *** markvoelker has joined #openstack-infra | 17:56 | |
| *** afazekas has quit IRC | 17:57 | |
| mgagne_ | I have a change with a -1 but no reviewer/CI with -1 in the reviewer list: https://review.openstack.org/#/c/171847/ can someone look into it? | 17:58 |
| *** mgagne_ is now known as mgagne | 17:58 | |
| *** dboik has quit IRC | 17:58 | |
| openstackgerrit | Jeremy Stanley proposed openstack-infra/reviewstats: Sync infra projects to governance repo list https://review.openstack.org/173021 | 17:59 |
| *** vinkman has left #openstack-infra | 17:59 | |
| *** annegentle has joined #openstack-infra | 17:59 | |
| *** otter768 has joined #openstack-infra | 18:00 | |
| *** pelix has quit IRC | 18:00 | |
| *** markvoelker has quit IRC | 18:01 | |
| zaro | mgagne: are you seeing -1 from rest request? | 18:01 |
| mgagne | zaro: I haven't debugged at this level =) | 18:02 |
| zaro | mgagne: where are you seeing the -1? i don't see it | 18:03 |
| fungi | mgagne: there was probably a verify -1 vote from a third-party ci which was later disabled from voting on that repo's changes, which causes the vote to no longer appear in the reviewers table of the change view but potentially leaves it in the dashboard list view | 18:03 |
| *** otter768 has quit IRC | 18:05 | |
| mgagne | fungi: how can this be fixed? my concern is that my change will now NOT be reviewed due to people filtering out changes with -1, all because a CI was voting -1 due to unrelated problems. | 18:05 |
| mgagne | fungi: getting something reviewed in Nova in a timely fashion is hard enough, now a defective CI made it even harder | 18:05 |
| AJaeger_ | mgagne: upload a new version, e.g. update the commit message | 18:05 |
| mgagne | AJaeger_: this shouldn't be the solution, shall I not asked here, I would have waited forever for a review | 18:06 |
| *** SumitNaiksatam has quit IRC | 18:07 | |
| fungi | mgagne: agreed. reviewers should not be skipping patches just because some random ci has left a -1 verify vote. if they do, it's that behavior which should be fixed | 18:07 |
| *** dangers is now known as dangers_away | 18:07 | |
| *** yfried has joined #openstack-infra | 18:07 | |
| mgagne | fungi: if a CI is disabled, can their votes be removed? | 18:07 |
| *** dangers_away is now known as dangers | 18:07 | |
| *** dangers is now known as dangers_away | 18:08 | |
| fungi | mgagne: not without the operators of that ci doing it themselves, or having a gerrit administrator spend time doing it manually (or using some tool which afaik does not yet exist) | 18:09 |
| mgagne | fungi: afaik, each project can disable a CI in a self-service way so I'm not sure if votes can programmatically be removed at the same time | 18:09 |
| fungi | zaro: is that ^ something someone has written for gerrit? (a tool which can delete votes when an account is removed from a group which would not have permitted them to leave that vote?) | 18:10 |
| mgagne | fungi: right. IMO, this should somehow be fixed as it leaves a very bad experience for the change owner | 18:10 |
| clarkb | so fwiw this is a bug in gerrit | 18:10 |
| fungi | mgagne: sounds like a great feature request, i agree https://code.google.com/p/gerrit/issues/list | 18:10 |
| clarkb | gerrit does attempt to remove the votes automagically | 18:10 |
| clarkb | it just fails at doing it properly | 18:10 |
| fungi | well, it attempts to hide them. it doesn't actually remove them from the database | 18:11 |
| *** dboik_ has quit IRC | 18:11 | |
| fungi | if the account's permissions are restored, any votes those permissions would have allowed them to leave will reappear | 18:11 |
| clarkb | sure, but more importantly it doesn't take them into account for merge calculations | 18:11 |
| mgagne | it sure does a great job at hiding them but still showing them in the review list | 18:11 |
| fungi | exactly. that part is definitely a gerrit bug | 18:11 |
| mgagne | there, I rebased... | 18:12 |
| zaro | fungi, mgagne : there's a change pending to remove reviewer votes: https://gerrit-review.googlesource.com/#/c/62410/ | 18:13 |
| *** sabeen has quit IRC | 18:13 | |
| mkj_wk | I see that there was discussion of dropping py26 testing -- would git-review be a special case, since the "py26" test is also testing that it works with git 1.7? | 18:13 |
| fungi | mkj_wk: we would probably keep that job around for as long as we keep centos6 as a job runner platform. that job is less a py26 test specifically, and more of a "does this work for centos/rhel 6.x test | 18:14 |
| mkj_wk | exactly | 18:14 |
| *** sushilkm has joined #openstack-infra | 18:14 | |
| mkj_wk | good to hear, even though the py26 job was the one that gave me the most grief :) | 18:14 |
| *** sushilkm has left #openstack-infra | 18:14 | |
| mkj_wk | (for that exact reason) | 18:15 |
| fungi | mkj_wk: the py26 dropping discussion was around libraries/clients which we were previously only releasing on their master branch but are now growing stable branches corresponding to our integrated release cycles | 18:15 |
| fungi | for those we can _potentially_ stop running py26 jobs on branches where py26 is no longer relevant for other projects | 18:16 |
| mkj_wk | yeah, just wanted to make sure git-review didn't suffer from a template change | 18:16 |
| mkj_wk | makes perfect sense | 18:16 |
| fungi | though that's still iffy for the clients themselves i think, and perhaps any of our libraries that those clients use | 18:16 |
| fungi | depends on whether we want to continue confirming that people can run latest python-novaclient on rhel/centos 6.x machines for example | 18:17 |
| clarkb | a redhat employee specifically writing using the non employee perspective said we should all stop making py2.6 work fwiw | 18:18 |
| clarkb | I think its fine to say git-review 1.x is the last release supporting py2.6 and stop testing py2.6 today | 18:18 |
| fungi | rhel/centos 7.x are a thing and have been for a while, after all | 18:18 |
| clarkb | then release 2.0 next release | 18:18 |
| fungi | which reminds me, we're overdue for another git-review release probably | 18:18 |
| *** sputnik13 has joined #openstack-infra | 18:19 | |
| mkj_wk | fungi: I'm just hoping my patch is integrated first :) | 18:19 |
| mkj_wk | fungi: folks here would like it | 18:19 |
| mkj_wk | most usage here is on py3 I think | 18:19 |
| *** cdent has quit IRC | 18:20 | |
| fungi | mkj_wk: yep, after some pretty rigorous manual testing that change seems to work fine for me (e.g. not break our cross-branch and cross-repo change-id duplicate workflows) | 18:21 |
| *** tlbr has quit IRC | 18:22 | |
| *** HeOS has joined #openstack-infra | 18:22 | |
| *** Sukhdev has quit IRC | 18:23 | |
| *** kgiusti has joined #openstack-infra | 18:23 | |
| *** teran has quit IRC | 18:24 | |
| *** AJaeger_ has quit IRC | 18:25 | |
| *** coolsvap|afk is now known as coolsvap | 18:26 | |
| *** aysyd has quit IRC | 18:28 | |
| *** aysyd has joined #openstack-infra | 18:28 | |
| *** isviridov has joined #openstack-infra | 18:28 | |
| *** ajmiller_ has joined #openstack-infra | 18:29 | |
| *** MaxV has joined #openstack-infra | 18:29 | |
| mkj_wk | mordred: if you are satisfied, a +2 on https://review.openstack.org/#/c/158877/ probably won't hurt. (I'm counting your rebasing on top of it as at least a small vote of confidence... ☺) | 18:31 |
| *** isviridov has quit IRC | 18:31 | |
| *** SlickNik has quit IRC | 18:31 | |
| *** SumitNaiksatam has joined #openstack-infra | 18:31 | |
| *** SlickNik has joined #openstack-infra | 18:31 | |
| mordred | mkj_wk: done! | 18:31 |
| *** isviridov_away is now known as isviridov | 18:32 | |
| mkj_wk | mordred: thanks! Do I need to do anything in the workflow to cause it to be merged? | 18:32 |
| *** annegentle has quit IRC | 18:33 | |
| *** ajmiller has quit IRC | 18:33 | |
| mordred | fungi: you good with me +Aing it? | 18:33 |
| fungi | mordred: yep | 18:33 |
| *** MaxV has quit IRC | 18:33 | |
| mordred | now I just have to get around to finsihing off my push-url patch | 18:35 |
| *** coolsvap is now known as coolsvap|afk | 18:36 | |
| *** tonytan_brb is now known as tonytan4ever | 18:36 | |
| mkj_wk | ☺ | 18:37 |
| *** tqtran has quit IRC | 18:38 | |
| openstackgerrit | Clark Boylan proposed openstack-infra/devstack-gate: Make multinode first class with ansible https://review.openstack.org/172614 | 18:38 |
| *** tqtran has joined #openstack-infra | 18:38 | |
| clarkb | fungi: sdague jeblair mordred et al ^ should be ready for review now | 18:38 |
| *** dizquierdo has quit IRC | 18:38 | |
| openstackgerrit | Jeremy Stanley proposed openstack-infra/project-config: Update gerritbot #openstack-infra from governance https://review.openstack.org/173032 | 18:38 |
| clarkb | the best part is it removes more code than it adds :) | 18:39 |
| fungi | clarkb: that's neat | 18:39 |
| fungi | what sort of speedup are you getting there? | 18:40 |
| clarkb | fungi: for the aiopcpu jobs is ~25 minutes | 18:40 |
| clarkb | maybe a little less than that | 18:40 |
| *** dboik has joined #openstack-infra | 18:40 | |
| fungi | that helps. how long does that get them down to now? | 18:40 |
| clarkb | fungi: its the same as non aiopcpu, ~40 minuts | 18:40 |
| *** armax has quit IRC | 18:40 | |
| fungi | ahh, okay | 18:41 |
| fungi | not bad! | 18:41 |
| clarkb | ya it has worked surprisingly well | 18:41 |
| clarkb | also, it resulted in some general d-g refactoring that I like, will probably do that if we decide not to ansible | 18:42 |
| *** ashleighfarnham has joined #openstack-infra | 18:42 | |
| *** armax has joined #openstack-infra | 18:42 | |
| clarkb | the big open question is whether or not openstack itself can handle being set up in parallel. Looks like the n-net job can, but I haven't confirmed that neutron can | 18:42 |
| clarkb | BUT if openstack can't I think we merge the changes then go fix openstack | 18:42 |
| openstackgerrit | Merged openstack-infra/git-review: Choose tracked branch for rebase when submitting https://review.openstack.org/158877 | 18:43 |
| mordred | clarkb: yeah - we can definitely make that a playbook or more playbook-like - but I agree, for right now just ansible commands in the shell script is the right choice | 18:43 |
| *** dprince has quit IRC | 18:43 | |
| clarkb | mordred: I have a few concerns so far with going more ansible like. | 18:43 |
| clarkb | mordred: the first is control flow. I think it works better in shell like this | 18:44 |
| clarkb | mordred: the second is ansible file manipulation is actually pretty :( | 18:44 |
| clarkb | mordred: you can't fetch a dir recursively for example, or reliably chown a dir recursively | 18:44 |
| mordred | clarkb: yah. | 18:44 |
| clarkb | mordred: or update a file by appending to it | 18:44 |
| mordred | clarkb: I agree | 18:44 |
| mordred | clarkb: this is why I've been happy about teh ansible/puppet split | 18:44 |
| anteaya | clarkb: you can update a file by appending to it | 18:45 |
| mordred | clarkb: ansible is not any cleaner than puppet once you get into config-management complexity sthings | 18:45 |
| clarkb | anteaya: how? | 18:45 |
| anteaya | or maybe not, this just ensures a line is in the file: http://docs.ansible.com/lineinfile_module.html | 18:45 |
| rcarrillocruz | insertafter EOF | 18:46 |
| rcarrillocruz | iirc | 18:46 |
| clarkb | rcarrillocruz: anteaya ya thats just a single line at a time | 18:46 |
| clarkb | what I want is to say take file X from src and append it to dest | 18:46 |
| clarkb | you can kinda do that with a template though | 18:47 |
| fungi | clarkb: nibalizer: did you have follow-up guidance for mrmartin on your comments in https://review.openstack.org/171600 ? | 18:47 |
| anteaya | the syncronize modules uses rsync under the hood | 18:47 |
| rcarrillocruz | http://docs.ansible.com/assemble_module.html ? | 18:47 |
| clarkb | fungi: mrmartin just copy the file? | 18:47 |
| anteaya | if you have a src file somewhere you can sync it to a dest | 18:47 |
| clarkb | rcarrillocruz: assemble is broken too | 18:47 |
| rcarrillocruz | use command: cat :P | 18:48 |
| clarkb | rcarrillocruz: because you have to have all the sourec files in one dir, and aiui you can't specify order | 18:48 |
| clarkb | rcarrillocruz: yes thats what I am saying, there is a lot of stuff that has to stay in bash | 18:48 |
| rcarrillocruz | yes | 18:48 |
| mordred | clarkb: you know, another thing we may want to play with is that ansible modules don't ahve to be in python - they can also be in shell or whatever | 18:48 |
| *** tkelsey has joined #openstack-infra | 18:48 | |
| rcarrillocruz | lineinfile is o-k for some stuff | 18:48 |
| mordred | clarkb: and ansible will handle copying them over and running them in that context | 18:48 |
| rcarrillocruz | i usually resort to templates | 18:48 |
| greghaynes | mordred: downside to that is youll have to mess with jq or something similar | 18:49 |
| clarkb | anyways if anyone wants to fix the related todays in my change feel free to push followup changes | 18:49 |
| mordred | greghaynes: only if you need to handle input params | 18:49 |
| clarkb | I think you will find that using bash is the simplest thing for many of them | 18:49 |
| * anteaya has to disembark the train soon | 18:49 | |
| mordred | clarkb: I'm not sure it's the right choice for anything - but it might be worth us making a synthetic example of using that mechanism so that we understand it | 18:49 |
| mordred | clarkb: ++ | 18:49 |
| mordred | I think using bash for shell scripting is a great choice | 18:49 |
| greghaynes | heh | 18:49 |
| clarkb | my biggest gripe is I can't fetch a dir | 18:49 |
| greghaynes | clarkb: synchronize? | 18:49 |
| mordred | and it we can use ansible help us stich that stuff together for cross-host execution in a m ore sane manner, sweet | 18:50 |
| clarkb | greghaynes: I think thats only ansible host -> rmote | 18:50 |
| clarkb | greghaynes: I need remote -> ansible host | 18:50 |
| greghaynes | I dont think so, its supposed to be all manner of direction | 18:50 |
| * clarkb goes to read more docs then | 18:50 | |
| greghaynes | thers a param to reverse direction IIRC | 18:50 |
| rcarrillocruz | my biggest gripe so far is the ssh agent forwarding with sudo | 18:51 |
| clarkb | greghaynes: gah its the dumbest param name ever "mode" | 18:51 |
| rcarrillocruz | you have to do some hackery to make it work | 18:51 |
| greghaynes | clarkb: haha, I was just scanning the docs thinking "I could have sworn there was one..." | 18:51 |
| greghaynes | so yes, name fail | 18:51 |
| nibalizer | fungi: looking | 18:51 |
| clarkb | greghaynes: thanks, i can use that | 18:51 |
| *** dhritishikhar_ has quit IRC | 18:52 | |
| fungi | clarkb: yeah, not sure whether you and nibalizer and mrmartin are talking past each other or not on that change. i'll dig into what the corresponding part of the solr module is doing to make that file available in the first place | 18:52 |
| clarkb | fungi: the problem is mrmartins change does not work | 18:52 |
| clarkb | fungi: so not sure why it is controversial to -1 | 18:52 |
| clarkb | we need to do something else, my suggestion is just a suggestion | 18:52 |
| fungi | after another reread of the commit message i think i see what he's trying to do. i'll see if i can test an alternative | 18:54 |
| *** adalbas has joined #openstack-infra | 18:54 | |
| nibalizer | fungi: cool that would be great | 18:56 |
| *** dannywilson has joined #openstack-infra | 18:57 | |
| fungi | wondering if it can just have the require on that file resource changed to the solr module (assuming it properly makes that source file present once it does all its things) | 18:57 |
| *** sayalilunkad has joined #openstack-infra | 18:57 | |
| *** hockeynut_ has quit IRC | 18:57 | |
| *** Rockyg has joined #openstack-infra | 18:58 | |
| *** hockeynut has joined #openstack-infra | 18:58 | |
| *** hashar has joined #openstack-infra | 18:59 | |
| *** dprince has joined #openstack-infra | 18:59 | |
| clarkb | mordred: oh the other ansible problem is environment | 18:59 |
| *** dboik has quit IRC | 19:00 | |
| clarkb | mordred: using the env in ansible is clunky at best. Probbaly not an issue once we zuulv3 since zuul can just pass in a json blob | 19:00 |
| openstackgerrit | David Shrewsbury proposed openstack-infra/shade: Add get_subnet() method https://review.openstack.org/171760 | 19:00 |
| *** ajmiller_ has quit IRC | 19:00 | |
| clarkb | but until we get there this practice of -m shell -a "source foo.sh && run thing executable=bash" is actually probably the simplest way to pass env around | 19:00 |
| clarkb | also the choice of /bin/sh as the shell executable is :( | 19:01 |
| mordred | clarkb: yah - passing env around from calling context is not really the intended mode of operation | 19:01 |
| *** Hal has joined #openstack-infra | 19:02 | |
| clarkb | mordred: which is unfortunate because this is the second time I have wanted it | 19:02 |
| clarkb | mordred: which makes me think either I am crazy or ansible is neglecting an important use case | 19:02 |
| mordred | clarkb: I think taht's because you need wanting to avoid writing things in python | 19:02 |
| *** Hal is now known as Guest97812 | 19:02 | |
| mordred | clarkb: I think it's doing what it's doing very much on purpose | 19:02 |
| mordred | and not on accident | 19:02 |
| rcarrillocruz | is it a known issue gate-infra-puppet-apply-precise test failing on exim upstart? | 19:02 |
| clarkb | mordred: I do not agree, writing python has little to do with this | 19:02 |
| clarkb | mordred: we want to execute shell on a remote host. Shells have environments | 19:03 |
| *** shashankhegde has quit IRC | 19:03 | |
| clarkb | ansible should be aware of that | 19:03 |
| mordred | no - it should very much not | 19:03 |
| rcarrillocruz | i keep on getting that error despite rechecks on https://review.openstack.org/#/c/172871/ | 19:03 |
| mordred | ansible is a tool for doing remote exec, it should NOT randomly pass the environment of the calling host to the remote host | 19:03 |
| mordred | that would be bonghits crazy in almost every case | 19:03 |
| clarkb | mordred: I don't want it to be random | 19:03 |
| clarkb | mordred: but there should be a method to do t | 19:04 |
| *** mrunge has quit IRC | 19:04 | |
| clarkb | mordred: one simple way is how puppet/facter do it. Prefix with FACTER_ | 19:04 |
| clarkb | mordred: anything with that prefix is essentially passed through | 19:04 |
| *** e0ne has joined #openstack-infra | 19:04 | |
| mordred | right - but the thing is that you're inventing a new way of doing complex action that is different from their method of complex action | 19:04 |
| clarkb | mordred: but their method of complex action for shells is terrible | 19:05 |
| mordred | instead of using the mechanism they have of chaining sets of actions together | 19:05 |
| clarkb | mordred: because it ignores the environment | 19:05 |
| mordred | nope | 19:05 |
| mordred | that's why I mentioned python | 19:05 |
| clarkb | rcarrillocruz: I do not think that is the error | 19:05 |
| *** sayalilunkad has quit IRC | 19:06 | |
| *** patrickeast_ has joined #openstack-infra | 19:06 | |
| rcarrillocruz | hmm | 19:06 |
| clarkb | rcarrillocruz: http://logs.openstack.org/71/172871/2/check/gate-infra-puppet-apply-precise/110b99d/console.html#_2015-04-13_11_27_30_171 I think that is the error | 19:06 |
| rcarrillocruz | first errors are puppet modules related | 19:06 |
| *** patrickeast has quit IRC | 19:06 | |
| *** patrickeast_ is now known as patrickeast | 19:06 | |
| clarkb | mordred: so to do anything with shell involving the env you think I as a user should write python? | 19:06 |
| fungi | rcarrillocruz: http://logs.openstack.org/71/172871/2/check/gate-infra-puppet-apply-precise/110b99d/console.html#_2015-04-13_11_27_30_171 | 19:06 |
| fungi | er, yep i agree with clarkb | 19:06 |
| rcarrillocruz | bleh | 19:07 |
| *** ajmiller has joined #openstack-infra | 19:07 | |
| rcarrillocruz | is it normal to have those upstart errors tho? | 19:07 |
| rcarrillocruz | kind of annoying to find a needle in a haystack | 19:07 |
| fungi | rcarrillocruz: the exim errors are because this is running in --noop and puppet is complaining that it can't restart things | 19:07 |
| clarkb | rcarrillocruz: yes, its running in noop mode so those init scripts don't make it onto the machine | 19:07 |
| rcarrillocruz | k, good to know | 19:07 |
| rcarrillocruz | thanks a bunch ! | 19:07 |
| clarkb | mordred: anyways, I think that any execution tool for unix running remote or not should have some env awareness | 19:08 |
| clarkb | mordred: because that is how processes on unix work | 19:08 |
| *** baoli_ has quit IRC | 19:09 | |
| bknudson | maybe this has been dealt with before? somebody has a pull request in github so somebody else proposes it to openstack? is that legal? | 19:09 |
| clarkb | they have an argv and and env and 3 sockets opened for them | 19:09 |
| fungi | rcarrillocruz: my guess is removing the instantiation of openstack_project::server in puppetdb.pp is causing your issue | 19:09 |
| clarkb | s/sockets/file descriptors/ | 19:09 |
| clarkb | bknudson: the answer is it depends | 19:09 |
| fungi | it's legal murky water | 19:10 |
| bknudson | seems like it. | 19:10 |
| clarkb | bknudson: whether or not its derivative, whether or not the submitter might actually own the code, whether or not the submitter has signed a CLA just submitted to the wrong place, etc all play into it | 19:10 |
| fungi | generally we try to encourage them to agree to the icla and propose the change themselves, or at least agree to the icla and let someone else propose the change | 19:10 |
| bswartz | fungi: there might be a problem with manila acls | 19:10 |
| bswartz | http://git.openstack.org/cgit/openstack-infra/project-config/tree/gerrit/acls/openstack/manila.config | 19:11 |
| bswartz | I see no mention of the proposed branch | 19:11 |
| bswartz | just stable | 19:11 |
| fungi | bswartz: did you create a branch? | 19:11 |
| mordred | clarkb: no - I think to do complex stuff you probably want to write python, and when you write complex things in python you probably want to parameterize it so that you can pass parameters in that make sense | 19:11 |
| bswartz | ttx created like an hour ago | 19:11 |
| fungi | bswartz: are you unable to do things on it? unspecified branches just inherit from refs/heads/* | 19:11 |
| mordred | clarkb: I think right now we're using it as a form of ssh, whcih is fine - but ssh doesn't pass invoking env do remote subprocesses | 19:11 |
| bswartz | no I haven't tried yet -- if it inherits then I foresee no issues | 19:12 |
| clarkb | mordred: actually :) I think it does | 19:12 |
| bswartz | just trying to avoid a fire drill if a backport should become necessary | 19:12 |
| clarkb | mordred: in some capacity, that is where the silly bash vulnerability from a while back got crazy | 19:12 |
| *** patrickeast has quit IRC | 19:12 | |
| clarkb | or maybe the network exposure was not via ssh? I thought ssh was involved | 19:13 |
| fungi | bswartz: usually the only reason to have a separate refs/heads/proposed/* section in the acl is if you're going to have a manila-milestone group with exclusive permissions on it | 19:13 |
| bswartz | I'll just test it to make sure | 19:13 |
| mordred | mordred@camelot:~$ ssh scylla.inaugust.com 'echo $DBUS_SESSION_BUS_ADDRESS' | 19:13 |
| mordred | that is empty | 19:13 |
| mordred | because ssh does not pass my local environment to the remote hose | 19:13 |
| mordred | host | 19:13 |
| mordred | I mean- I know what you want to do | 19:13 |
| clarkb | mordred: ah its configurable | 19:13 |
| mordred | but I thin you're trying to get one tool to seamlessly work with another tool when you're using one of them only halfway | 19:14 |
| clarkb | mordred: you need to PermitUserEnvironment sshd side then client side select which to pass through | 19:14 |
| mordred | yah | 19:14 |
| clarkb | mordred: and really something like ^ is all I ask ansible to do | 19:14 |
| mordred | and if you select which things to pass through | 19:14 |
| *** e0ne has quit IRC | 19:14 | |
| mordred | it HAS that | 19:14 |
| clarkb | mordred: HOW? | 19:14 |
| mordred | it's call passing parameters to a module | 19:14 |
| fungi | bswartz: for that matter the refs/heads/stable/* section in there is redundant now that it's been redelegated to manila-core. if that section were omitted it would just inherit the same things from refs/heads/* now that we no longer have a refs/heads/stable/* in the All-Projects global acl with exclusive group permissions for the stable branch managers | 19:14 |
| clarkb | mordred: but how do they get into the module from the remote side? | 19:14 |
| mordred | you pass them in as parameters at the invoking side | 19:15 |
| mordred | and they are passed over the module invocation interface | 19:15 |
| clarkb | mordred: but how do I get the info from the remote side? | 19:15 |
| mordred | what? | 19:15 |
| clarkb | mordred: I need the env on $HOSTOVERTHERE | 19:15 |
| clarkb | mordred: but I am runnin ansible of localhost | 19:15 |
| mordred | you want to read the env of HOSTOVERTHERE | 19:16 |
| clarkb | mordred: yes | 19:16 |
| mordred | that's easy | 19:16 |
| mordred | all of yoru scripts are running on that host | 19:16 |
| mordred | so, they can just read the envionment | 19:16 |
| clarkb | mordred: right so my current workaround is to do that | 19:16 |
| clarkb | mordred: but its clunky | 19:16 |
| mordred | of course its clunky | 19:16 |
| mordred | you're not really using the tool the way it's designed | 19:16 |
| mordred | so one might assume some clunk :) | 19:16 |
| clarkb | I guess I don't understand how its designed if not to remotely run code | 19:17 |
| clarkb | remotely run code needs an env | 19:17 |
| mordred | it's not designed to do complex actions through ad-hoc command line invocations | 19:17 |
| clarkb | because processes on unix have environments | 19:17 |
| mordred | wait - can you describe your problem again | 19:17 |
| *** e0ne has joined #openstack-infra | 19:17 | |
| mordred | I think I'm COMPLETELY missing what's bothering you | 19:17 |
| mordred | you have host and target, and each has an env, host-env and target-env | 19:18 |
| mordred | what is where that you are missing from where? | 19:18 |
| clarkb | mordred: the problem is that I want ansible to be aware of remote execution environments when it runs. So that I don't have to do -m shell -a "source foo.sh && do something executable=bash"" | 19:18 |
| mordred | clarkb: what's in foo.sh and why is it there? | 19:18 |
| clarkb | mordred: I want to do -m command -a "do stuff env=somehowdescribeenv" | 19:18 |
| clarkb | mordred: that is what defines the remote env | 19:18 |
| mordred | clarkb: what created the contents of it? | 19:18 |
| clarkb | mordred: in this case devstack-gate | 19:18 |
| mordred | right | 19:19 |
| mordred | so - this is what I was saying earlier | 19:19 |
| mordred | you have a set of parameters in d-g on tehe host side | 19:19 |
| mordred | and you want to get some of those to -a "do stuff" | 19:19 |
| clarkb | mordred: its not just host side | 19:19 |
| clarkb | mordred: its host and remote side both | 19:19 |
| mordred | you have two different environments and you want to mix them | 19:19 |
| clarkb | yes | 19:19 |
| *** marun has quit IRC | 19:20 | |
| mordred | how do you expect ansible to know which env vars to pass from host to target? | 19:20 |
| clarkb | even the puppet method of FACTER_ prefixed env vars would work just fine | 19:20 |
| *** marun has joined #openstack-infra | 19:20 | |
| mordred | you could do that just fine | 19:20 |
| mordred | but I'm really trying to undersatnd specifically what you want because it makes no sense to me and I want it to | 19:20 |
| mordred | then I can help suggest how to do it - or a patch to upstream :) | 19:21 |
| clarkb | mordred: I want to avoid the shell module for two reasons, its less secure and harder to reason about as a result and its only useful if I override the shell executable to be bash | 19:21 |
| *** coolsvap|afk is now known as coolsvap | 19:21 | |
| clarkb | mordred: instead I should be able to use the command module, except that unix commands needs environments to operate properly and ansible fails at this | 19:21 |
| mordred | AH | 19:21 |
| mordred | so - you want to not run things in a subshell, but you also want for unix environments to work | 19:22 |
| clarkb | yes | 19:22 |
| mordred | I think literally no tool is ever going to do that for you transparently - but I will work on a follow up patch to yoru patch for a sake-of-argument maybe-this-is-closer-to-what-you-want | 19:22 |
| mordred | I believe I grok the thing you want now | 19:23 |
| clarkb | it doesn't need to be transparent | 19:23 |
| mordred | sure | 19:23 |
| mordred | no - I get it | 19:23 |
| mordred | let me take some code stabs | 19:23 |
| clarkb | if I can say pass through all of the DEVSTACK_GATE vars from host to remote and read remote env X | 19:23 |
| clarkb | that would work fine | 19:23 |
| *** carl_baldwin_ has quit IRC | 19:24 | |
| clarkb | mordred: you will notice that I have decided to do all variable interpolation on the local ansible machine | 19:24 |
| clarkb | mordred: this is an incomplete attempt at making the use of shell not super terrible | 19:25 |
| clarkb | its not properly escaping everything so could be used to do weirdness on the remote side | 19:25 |
| clarkb | but at least it should be logged properly on the localhost side that its doing so before it does that | 19:25 |
| clarkb | we will not be able to do ^ that in zuulv3 | 19:26 |
| clarkb | zuulv3 will need remote resolution of environment to work sanely | 19:27 |
| *** sabeen has joined #openstack-infra | 19:27 | |
| *** sabeen has quit IRC | 19:27 | |
| *** dboik has joined #openstack-infra | 19:28 | |
| *** Sukhdev has joined #openstack-infra | 19:28 | |
| clarkb | or we can not use command and instead use shell whenever we need to construct an environment | 19:29 |
| mordred | I believe that is what you need to do | 19:29 |
| mordred | command explicitly does not construct an environment | 19:30 |
| openstackgerrit | Khai Do proposed openstack-infra/system-config: update Gerrit javamelody plugin on review-dev.o.o https://review.openstack.org/173050 | 19:30 |
| sdague | clarkb: so, stack.sh in parallel is not recommended | 19:31 |
| *** soren has joined #openstack-infra | 19:31 | |
| sdague | reed: https://review.openstack.org/#/c/172614/16 | 19:31 |
| clarkb | sdague: can you expand on that? | 19:31 |
| clarkb | sdague: because stack.sh shouldn't care | 19:31 |
| sdague | well, it means you can get a nova-compute up on a node before rabbit is working, for instance | 19:32 |
| mordred | clarkb: but I'll hack up some options that hopefully suck less | 19:32 |
| clarkb | sdague: yes, that is the case, but nova compute handles that fine | 19:32 |
| sdague | some times | 19:32 |
| *** kmartin has quit IRC | 19:33 | |
| sdague | it means the service check in for nova compute on boot isn't going to happen, which might mean goofiness with the scheduler | 19:33 |
| sdague | anyway, there are a lot of potential gotchas here | 19:33 |
| *** xyang1 has joined #openstack-infra | 19:33 | |
| fungi | clarkb: nibalizer: so after playing around some more with 171600 i see the problem. the source for that file is short-lived, and puppet expects it to stick around. we want to make sure the file is copied to where the application expects it if it doesn't exist yet, but not bank on the source of the file sticking around forever | 19:33 |
| clarkb | sdague: would it be better to run in parallel and cut 20 minutes off these jobs and fix the problems as we find them or run in serial and be made at devstack for being so slow? | 19:33 |
| fungi | clarkb: nibalizer: and we don't control the module which has chosen /tmp as the place to unpact that file | 19:34 |
| clarkb | sdague: /me is trying to figure out where I should best focus my energy | 19:34 |
| fungi | er, unpack | 19:34 |
| sdague | clarkb: subnode setup is 20 minutes? | 19:34 |
| clarkb | sdague: yes | 19:34 |
| sdague | can you point me to that fact? | 19:34 |
| clarkb | sdague: its about 5 minutes for workspace setup and 15 for devstack | 19:34 |
| openstackgerrit | Khai Do proposed openstack-infra/system-config: update Gerrit javamelody plugin on review.o.o https://review.openstack.org/173052 | 19:34 |
| clarkb | sdague: ya I can get logs to compare | 19:35 |
| fungi | clarkb: nibalizer: i suppose an alternative is to manually fix it up by copying that source file back so that puppet stops complaining, and then disable the tmpreaper on the server or at least whitelist the directory where solr unpacks itself? | 19:35 |
| sdague | like logs, because total devstack setup is about 15 minutes, but 4 of that is keystone starting and making accounts | 19:35 |
| *** hdd has quit IRC | 19:35 | |
| sdague | which is no way should be in the subnode | 19:35 |
| sdague | I did look at that, especially around the osc overhead (see ML post this morning) | 19:35 |
| clarkb | sdague: http://logs.openstack.org/65/172465/2/check/check-tempest-dsvm-aiopcpu-full/6f4611e/console.html thats a serial run | 19:36 |
| clarkb | took ~61 minutes | 19:37 |
| clarkb | sdague: http://logs.openstack.org/14/172614/13/experimental/check-tempest-dsvm-aiopcpu-full/599095f/console.html parallel run took ~43 minutes | 19:37 |
| sdague | clarkb: right, subnode stack.sh is 3 minutes - http://logs.openstack.org/65/172465/2/check/check-tempest-dsvm-aiopcpu-full/6f4611e//logs/10.209.67.74-subnode/devstacklog.txt.gz | 19:38 |
| clarkb | hrm how is the parallel run so much faster then? | 19:39 |
| *** teran has joined #openstack-infra | 19:39 | |
| clarkb | we only save ~5 minutes in host setup and workspace setup | 19:40 |
| clarkb | ~3 minutes from stack.sh so thats a total of 8 minutes. Currently about 10 minutes short | 19:40 |
| *** bradjones has quit IRC | 19:40 | |
| clarkb | looks like the normal stack.sh on the serial run took almost 4 minutes longer | 19:41 |
| clarkb | so that may just be skew | 19:41 |
| clarkb | sdague: also can you explain "the service check in for nova compute on boot isn't going to happen" | 19:41 |
| sdague | clarkb: tempest took an extra 5 minutes in the longer one (all of tempest ran slower it looks like), not sure wwy | 19:41 |
| *** flashgordon has quit IRC | 19:42 | |
| clarkb | we don't reboot any of these nodes, so no sure what boot time checks we would be making | 19:43 |
| sdague | nova compute does a lot of things in init that assume the rest of the services are up. There is a reason we start the nova services in a specific order | 19:43 |
| *** isviridov is now known as isviridov_away | 19:43 | |
| clarkb | sdague: and we don't consider that a bug with nova? we have to restart all computes if the controller node is rebooted? | 19:44 |
| clarkb | anyways new patch coming shortly that splits out stack.sh runs | 19:44 |
| sdague | the restart case is different, because some of this is first time data initialization in the database | 19:44 |
| *** emagana has joined #openstack-infra | 19:46 | |
| *** shashankhegde has joined #openstack-infra | 19:47 | |
| openstackgerrit | Clark Boylan proposed openstack-infra/devstack-gate: Make multinode first class with ansible https://review.openstack.org/172614 | 19:47 |
| openstackgerrit | Doug Hellmann proposed openstack-infra/project-config: Add python3 jobs to oslo.messaging https://review.openstack.org/173056 | 19:47 |
| *** dboik has quit IRC | 19:48 | |
| *** cdent has joined #openstack-infra | 19:48 | |
| *** coolsvap is now known as coolsvap|afk | 19:48 | |
| sdague | it also eventually heals on the periodic jobs, but eventually might not be good enough here. So I'd rather play it a little safe here, especially as the time cost is only 3 minutes. | 19:49 |
| clarkb | sdague: fwiw I think openstack services like nova compute, neutron l3, neutron agent, etc, basically anything expected to run off controller should be able to come up without distress before the controller | 19:49 |
| sdague | clarkb: they can, it doesn't mean they'll be in the cluster | 19:50 |
| clarkb | right, openstack should fix that | 19:50 |
| sdague | like I said, it heals eventually in periodic | 19:50 |
| sdague | it's eventually consistent, which is fine for most cases | 19:50 |
| clarkb | maybe sighup to reconnect to cluster | 19:51 |
| sdague | it really is kind of an edge case though, it's like saying "you should totally be able to start api servers with no db" | 19:52 |
| fungi | dhellmann: does https://review.openstack.org/173056 mean that the bug with oslo.messaging unit tests on the python 3.4 in ubuntu trusty is no longer a problem? | 19:53 |
| *** sdake has joined #openstack-infra | 19:53 | |
| clarkb | sdague: I don't think its an edge case, losing your control plane is something that happens frequently | 19:53 |
| sdague | it's not about losing the control plane | 19:54 |
| *** patrickeast has joined #openstack-infra | 19:54 | |
| sdague | it's about it never existing | 19:54 |
| clarkb | sdague: ok, I guess I still don't understand then | 19:54 |
| sdague | nova-compute comes up | 19:54 |
| clarkb | sdague: these are not effectively the same issue? either its not there or it is there | 19:54 |
| *** kmartin has joined #openstack-infra | 19:54 | |
| clarkb | and whether or not it previously existed shouldn't affect that | 19:55 |
| sdague | so... what's the mechanism for a nova-compute to tell the cluster it exists? | 19:55 |
| clarkb | sdague: rabbit | 19:55 |
| sdague | before that | 19:55 |
| sdague | when does nova-compute tell the world it exists? | 19:55 |
| sdague | what event causes that to happen | 19:55 |
| clarkb | sdague: when it comes up, then it tries again on a poll and hopefully on a sighup | 19:55 |
| nibalizer | fungi: so i think the real solution is to use this: https://github.com/nanliu/puppet-archive#resources | 19:56 |
| sdague | well, the answer is, when it boots, and on a periodic cycle | 19:56 |
| nibalizer | that puppet extension allows you to say 'put this file here, but get it out of a tarball from $HTTP_URL' | 19:56 |
| clarkb | sdague: sure so that seems fine | 19:56 |
| *** sdake_ has quit IRC | 19:56 | |
| nibalizer | now honestly, i've not had time to bang on this particular issue and we'd have to patch heavily the upstream solr module, or roll our own | 19:56 |
| *** achanda has quit IRC | 19:57 | |
| fungi | nibalizer: use it where? the module which is retrieving and extracting that tarball is one we didn't write | 19:57 |
| mordred | nibalizer: isn't there a puppet module that's being used in the storyboard modules to do a similar thing? | 19:57 |
| *** coolsvap has joined #openstack-infra | 19:57 | |
| sdague | but the "when it boots" doesn't just hang until there is an answer, it tries, and moves on figuring it can be consistent later (IIRC) | 19:57 |
| fungi | nibalizer: though i guess we could guess at the location where it stashes the tarball (if it keeps it somewhere better than /tmp hopefully?) and then reextract it to get that file out even though we don't need it because we already have it in the right destination | 19:57 |
| sdague | so my concern is you end up trying to do test work in the gate between booting/ failing to declare itself, and the first periodic checkin after | 19:58 |
| clarkb | sdague: right which is why I suggest a sighup | 19:58 |
| clarkb | (or similar mechanism) | 19:58 |
| sdague | ok, well that's code that doesn't exist today | 19:58 |
| clarkb | becuse we hopefully aren't the only people that want to make sure their cluster synchronizes after a reboot | 19:58 |
| openstackgerrit | Khai Do proposed openstack-infra/puppet-gerrit: Update bouncy castle libs for Gerrit 2.10 https://review.openstack.org/172534 | 19:59 |
| openstackgerrit | Khai Do proposed openstack-infra/puppet-gerrit: Update bouncy castle libs for Gerrit 2.10 https://review.openstack.org/172534 | 19:59 |
| sdague | clarkb: it does after a reboot, because nova-compute ran init_host again | 19:59 |
| nibalizer | mordred: there is 'example42/puppi' in the metadata.json | 19:59 |
| *** annegentle has joined #openstack-infra | 19:59 | |
| nibalizer | i dont see it actually used though | 19:59 |
| nibalizer | i try to stay far far away from anything example42 writes | 19:59 |
| sdague | it's about the record ever existing in the first place | 19:59 |
| clarkb | sdague: but only if you reboot the compute side | 19:59 |
| sdague | clarkb: no no no no no | 20:00 |
| nibalizer | fungi: yea we'd have to send that patch upstream, and they may or may not be amenable to that | 20:00 |
| clarkb | sdague: then how does compute side know to recheck back in outside of poll interval otherwise? | 20:00 |
| sdague | clarkb: it doesn't, and it's not important | 20:00 |
| openstackgerrit | Timo Tijhof proposed openstack-infra/zuul: Status: Don't raise "abort" as error to the user https://review.openstack.org/172241 | 20:00 |
| clarkb | sdague: ok then I am missing what the problem is still | 20:00 |
| sdague | because it registered itself into the nova service list once | 20:00 |
| *** dboik has joined #openstack-infra | 20:01 | |
| sdague | so the scheduler will account for it assuming it's up | 20:01 |
| *** otter768 has joined #openstack-infra | 20:01 | |
| dhellmann | fungi: oh, no, it means I forgot about that. :-/ | 20:01 |
| sdague | and if it finds out later it's down, then it will mark it down until the compute tells it it's back up | 20:01 |
| fungi | nibalizer: well, part of the trick is i think that we're wanting something out of the tarball which that other puppet module isn't installing. better would be to just upstream a patch to add an option to tell it to install that additional file from its tarball | 20:01 |
| sdague | but you are starting from a blank world | 20:01 |
| nibalizer | fungi: yes if the tarball is stored in a permament location we could do what you're saying | 20:01 |
| nibalizer | fungi: well 'better' | 20:01 |
| clarkb | sdague: but blank state is effectively down too | 20:01 |
| nibalizer | certainly the patch you describe would be easier to land | 20:02 |
| dhellmann | fungi: well, it worked here on my box, too. Is there some relatively easy way to reproduce that failure? | 20:02 |
| clarkb | sdague: so in both cases you just need something that tells nova to say "I'm up" | 20:02 |
| *** Guest97812 has quit IRC | 20:02 | |
| fungi | dhellmann: i can try to 'tox -e py27' in a recent checkout on an up to date ubuntu trusty vm | 20:02 |
| nibalizer | but i've been here before with puppet + tarballs + wget + execs and it never gets better it just grows in complexity until its unmanagable/super racy and you throw it in the trash | 20:02 |
| sdague | yep, and it's called starting the nova service ... or waiting 10 minutes | 20:02 |
| clarkb | sdague: it only waits 10 seconds fwiw | 20:02 |
| dhellmann | fungi: ok, I did that on my dev box but it may not be up to date | 20:02 |
| dhellmann | fungi: I'll be more explicit | 20:02 |
| clarkb | oh hrm wait | 20:02 |
| sdague | clarkb: I was pretty sure we tuned that down | 20:03 |
| clarkb | sdague: so nova does two polls, the rabbit poll is every 5 seconds. The conductor poll is every 10 seconds then it does a backoff | 20:03 |
| clarkb | thats not confusing at all :/ | 20:03 |
| *** cdent has quit IRC | 20:03 | |
| sdague | clarkb: yeh, but neither of those is actually the thing | 20:03 |
| fungi | dhellmann: it's bug 1367907 | 20:03 |
| openstack | bug 1367907 in python3.4 (Ubuntu Trusty) "Segfault in gc with cyclic trash" [Undecided,Triaged] https://launchpad.net/bugs/1367907 | 20:03 |
| mordred | nibalizer, fungi: AFS | 20:04 |
| clarkb | sdague: according to the logs that is what kicks off the config portion of nova compute | 20:04 |
| dhellmann | fungi: thanks | 20:04 |
| clarkb | sdague: once it can check in with conductor the service starts running | 20:04 |
| sdague | clarkb: ok, if it polls forever, that's probably fine. I didnt' realize it did that | 20:04 |
| clarkb | sdague: then once config is read in it does the registration you are worried about | 20:04 |
| sdague | I thought we stopped after some period of time | 20:05 |
| dhellmann | who is the gerrit acl file expert? I have a question about configuring the oslo acls to allow core teams to be the same as stable teams without maintaining separate groups in gerrit | 20:05 |
| fungi | dhellmann: deleting the refs/heads/stable/* section from your acls ought to do it | 20:05 |
| *** imcsk8|afk is now known as imcsk8 | 20:05 | |
| clarkb | sdague: I am looking at http://logs.openstack.org/14/172614/13/experimental/check-tempest-dsvm-aiopcpu-full/599095f/logs/10.209.0.151-subnode/screen-n-cpu.txt.gz fwiw | 20:05 |
| dhellmann | fungi: excellent, that's much easier than what I was doing :-) | 20:05 |
| *** otter768 has quit IRC | 20:05 | |
| fungi | dhellmann: alternatively you could have those separate teams just include the core team and the stable maint team | 20:06 |
| fungi | dhellmann: that is if you want openstack stable maint to also be pitching in as core reviewers on those | 20:06 |
| sdague | clarkb: ok, I didn't realize we'd block until then | 20:07 |
| sdague | I thought it gave up at some point | 20:07 |
| dhellmann | fungi: oh, hrm, that's a fair point | 20:07 |
| *** e0ne has quit IRC | 20:08 | |
| dhellmann | fungi: right now we just have oslo-stable-maint with one or two exceptions, I think | 20:08 |
| fungi | dhellmann: we could also tweak it so that the refs/heads/stable/* section doesn't have an exclusiveGroupPermissions line, and then add the stable maint team in place of the core team there, and let it be additive to what it inherits from refs/heads/* | 20:08 |
| clarkb | sdague: anyways its only 3 minutes. I am happy to use the latest patchset which always brings up controller first then subnodes. Then later we can worry about correctness of removing that 3 minute cost | 20:08 |
| sdague | clarkb: ok, cool | 20:08 |
| *** dannywilson has quit IRC | 20:08 | |
| clarkb | it does do all subnodes in parallel so that if we run on >2 nodes we see a cost savings | 20:09 |
| openstackgerrit | Michael Krotscheck proposed openstack-infra/puppet-storyboard: Cron configuration renamed to [scheduler] https://review.openstack.org/173063 | 20:09 |
| openstackgerrit | Michael Krotscheck proposed openstack-infra/puppet-storyboard: Enabled Cron Token Scheduler https://review.openstack.org/173064 | 20:09 |
| sdague | yeh, subnodes in parallel makes total sense to me | 20:09 |
| *** annegentle has quit IRC | 20:10 | |
| sdague | I'm also a little concerned that even though it functionally recovers, it leaves all those errors (which are correct errors) which are going to mask out them showing up in actual failures | 20:10 |
| dhellmann | fungi: and the stable maint team is "stable-maint"? | 20:10 |
| dhellmann | no, stable-maint-core | 20:11 |
| fungi | dhellmann: yep, the latter https://review.openstack.org/#/admin/groups/530,members | 20:11 |
| dhellmann | fungi: ok, that sounds like what I want then, I'll do that. Thanks! | 20:11 |
| *** asettle is now known as asettle-gym | 20:11 | |
| *** hdd has joined #openstack-infra | 20:13 | |
| dhellmann | fungi: haypo pointed out that my py3 patch was a dupe of https://review.openstack.org/#/c/172221/ | 20:13 |
| *** tlbr has joined #openstack-infra | 20:14 | |
| fungi | dhellmann: aha, excellent, and it also links to the change to make it work | 20:15 |
| fungi | i'll manually confirm it since i've held a machine for that purpose anyway | 20:15 |
| clarkb | vivid has 3.4.3 in it, how do we find out if trusty is going to get that? | 20:15 |
| fungi | clarkb: we rattle cages at canonical again apparently | 20:16 |
| fungi | dhellmann: even without https://review.openstack.org/172135 i'm no longer getting a failure for tox -e py34 on a held bare-trusty worker now | 20:17 |
| fungi | so something else may have caused that bug to go back into hiding | 20:17 |
| mtreinish | clarkb: ooh 3.4.3, finally the same version my boxes default too | 20:19 |
| dhellmann | fungi: yay? :-) | 20:22 |
| fungi | dhellmann: i'm retrying a few more times just to be sure, but seems that way | 20:23 |
| dhellmann | fungi: excellent | 20:24 |
| fungi | worrisome. the bugs that hide are much more aggravating that those which dangle right in front of you | 20:25 |
| fungi | oh, neat, seems there's a race in mox though | 20:25 |
| *** achanda has joined #openstack-infra | 20:25 | |
| *** ftersin has joined #openstack-infra | 20:28 | |
| mordred | clarkb: wendar was going to poke doko at pycon about 3.4.3 | 20:29 |
| mordred | clarkb: or at least about unbreaking 3.4 | 20:29 |
| *** dizquierdo has joined #openstack-infra | 20:31 | |
| *** yfried is now known as yfried|afk | 20:32 | |
| *** jgrimm is now known as zz_jgrimm | 20:33 | |
| *** HenryG has quit IRC | 20:33 | |
| *** kmartin has quit IRC | 20:34 | |
| *** dkranz has quit IRC | 20:35 | |
| *** yfried|afk is now known as yfried | 20:37 | |
| *** HenryG has joined #openstack-infra | 20:37 | |
| openstackgerrit | Doug Hellmann proposed openstack-infra/project-config: Use Oslo core teams for stable maint, too https://review.openstack.org/173075 | 20:38 |
| *** baoli has joined #openstack-infra | 20:38 | |
| dhellmann | fungi: I wanted to include some comments in ^^ but I can't find any examples of a comment syntax in acl files. | 20:38 |
| *** hdd has quit IRC | 20:38 | |
| openstackgerrit | Julia Kreger proposed openstack-infra/shade: Add update_machine method and operator unit test substrate https://review.openstack.org/171165 | 20:41 |
| openstackgerrit | Julia Kreger proposed openstack-infra/shade: Update register_machine to use tasks https://review.openstack.org/171324 | 20:41 |
| openstackgerrit | Julia Kreger proposed openstack-infra/shade: Update unregister_machine to use tasks https://review.openstack.org/171325 | 20:41 |
| *** e0ne has joined #openstack-infra | 20:41 | |
| *** radez is now known as radez_g0n3 | 20:41 | |
| *** nbernard has joined #openstack-infra | 20:42 | |
| *** baoli_ has joined #openstack-infra | 20:45 | |
| *** unicell has quit IRC | 20:45 | |
| clarkb | mordred: yay! | 20:45 |
| *** peristeri has quit IRC | 20:46 | |
| *** yfried is now known as yfried|afk | 20:47 | |
| *** dkranz has joined #openstack-infra | 20:47 | |
| *** baoli has quit IRC | 20:47 | |
| wendar | mordred: we've got a debian python bof with doko tomorrow at sprints, will poke him then | 20:48 |
| wendar | mordred: (had dinner with him last night, but seemed like a topic that needs a laptop) | 20:49 |
| openstackgerrit | Merged openstack-infra/shade: Rename auth_plugin to auth_type https://review.openstack.org/171432 | 20:50 |
| *** ashleighfarnham has quit IRC | 20:50 | |
| *** ajo has quit IRC | 20:51 | |
| *** kmartin has joined #openstack-infra | 20:51 | |
| openstackgerrit | Clark Boylan proposed openstack-infra/system-config: Ensure latest on etherpad vcsrepo https://review.openstack.org/164323 | 20:52 |
| clarkb | mordred: fungi ^ | 20:52 |
| *** cference_ has quit IRC | 20:52 | |
| clarkb | I will do a db dump prior to that getting applied so either let me know if you are going to approve it or just +2 and I can approve | 20:52 |
| *** sdake_ has joined #openstack-infra | 20:53 | |
| openstackgerrit | Julia Kreger proposed openstack-infra/shade: Add test of OperatorCloud auth_type=None https://review.openstack.org/172165 | 20:54 |
| *** tiswanso_ has quit IRC | 20:54 | |
| *** andreykurilin has quit IRC | 20:55 | |
| *** rbradfor_ has joined #openstack-infra | 20:56 | |
| *** hichihara has joined #openstack-infra | 20:56 | |
| *** sdake has quit IRC | 20:57 | |
| krotscheck | greghaynes: Hey, did that pymysql oslo.db error ever get fixed? | 20:57 |
| krotscheck | greghaynes: i.e. is it in a newer version of oslo.db? | 20:58 |
| clarkb | krotscheck: I think it did https://review.openstack.org/#/c/155574/ that merged back in february and oslo.db has been released since then | 20:58 |
| *** Rockyg has quit IRC | 20:58 | |
| * clarkb is double checking on release time | 20:58 | |
| krotscheck | clarkb: neat | 20:58 |
| *** markvoelker has joined #openstack-infra | 20:59 | |
| clarkb | 1.7.1 was in march and we just had a release today | 20:59 |
| *** marcusvrn has quit IRC | 20:59 | |
| openstackgerrit | Julia Kreger proposed openstack-infra/shade: Add test of OperatorCloud auth_type=None https://review.openstack.org/172165 | 21:00 |
| krotscheck | clarkb: grazie, that's what I needed. | 21:01 |
| *** dimtruck is now known as zz_dimtruck | 21:01 | |
| *** tkelsey has quit IRC | 21:02 | |
| *** hdd has joined #openstack-infra | 21:03 | |
| greghaynes | krotscheck: yep | 21:03 |
| greghaynes | ah, clarkb go tit | 21:03 |
| *** gmurphy has left #openstack-infra | 21:03 | |
| *** markvoelker has quit IRC | 21:03 | |
| fungi | dhellmann: close, just one more simplification you can make there | 21:04 |
| *** aysyd has quit IRC | 21:07 | |
| fungi | clarkb: what does ensure latest do with a specific revision in vcsrepo? those seem somewhat contradictory | 21:08 |
| clarkb | fungi: hopefully I explained that in the commit message | 21:08 |
| clarkb | fungi: if you don't ensure latest the vcsrepo doesn't update anything | 21:08 |
| fungi | you know what? you totally did and i went straight to the diff :/ | 21:08 |
| clarkb | its ok, gerrit makes that too easy :) | 21:08 |
| fungi | indeed | 21:08 |
| *** annegentle has joined #openstack-infra | 21:09 | |
| *** cdent has joined #openstack-infra | 21:09 | |
| fungi | clarkb: lgtm | 21:09 |
| krotscheck | greghaynes: Just for sanity's sake, if I pull these except catches out of the codebase, the unit tests should pick up errors in an older version of oslo, yes? | 21:10 |
| *** mmedvede has quit IRC | 21:11 | |
| fungi | dhellmann: i'm not getting that mox failure on oslo.messaging py34 again. seems to maybe be somewhat rare. i got it on my second run but not in the several since | 21:11 |
| *** aysyd has joined #openstack-infra | 21:11 | |
| fungi | not sure if that's py3k-specific or just so rare you're not noticing it on 2.x either | 21:12 |
| dhellmann | fungi: sounds like a race, so it's probably not related to python 3, unless there's some difference in behavior when eventlet is not used | 21:12 |
| dhellmann | yep | 21:12 |
| greghaynes | krotscheck: where in the code? | 21:13 |
| *** stevemar has quit IRC | 21:13 | |
| hashar | fungi: you might want to tag a new version of nodepool since https://review.openstack.org/#/c/170289/ changed the mysql module being used :) | 21:13 |
| krotscheck | greghaynes: http://git.openstack.org/cgit/openstack-infra/storyboard/tree/storyboard/db/api/base.py#n287 | 21:14 |
| krotscheck | greghaynes: There's two more in there. | 21:14 |
| *** ZZelle_ has quit IRC | 21:14 | |
| *** dizquierdo has quit IRC | 21:15 | |
| fungi | hashar: sure, can do. do we need a 0.0.2 just before the dependency change and then a 0.0.3 after? | 21:15 |
| *** ldnunes has quit IRC | 21:15 | |
| fungi | hashar: or should i tag that commit 0.0.2? | 21:15 |
| fungi | the couple of commits which have merged since seem benign enough to just wrap up into 0.0.2 if you're not particularly picky | 21:16 |
| *** hdd has quit IRC | 21:16 | |
| greghaynes | krotscheck: ah, yes. that is a gross workaround | 21:17 |
| greghaynes | that I wrote | 21:17 |
| fungi | hashar: i can put it on the meeting agenda for tomorrow and then take care of it once others have a chance to object to tagging a new release for it | 21:17 |
| krotscheck | greghaynes: Right, but the awesome fix you put in oslo_db is now available, so I'm trying to figure out how to roll back that hack. | 21:17 |
| greghaynes | krotscheck: You could just remove that last except block, the issue is that techincally you wouldnt support the versions of oslo.db you have in your requirements then | 21:18 |
| krotscheck | greghaynes: That's part of a different pathc | 21:19 |
| *** tnovacik has quit IRC | 21:19 | |
| krotscheck | greghaynes: I'll add you to the review when it goes up. | 21:19 |
| krotscheck | greghaynes: It looks like you removed a DBReferenceError catch earlier. | 21:19 |
| greghaynes | ok | 21:20 |
| *** andreykurilin has joined #openstack-infra | 21:20 | |
| clarkb | mordred: can I get a review on 164323 again? if you +2 I can approve and babysit | 21:20 |
| clarkb | mordred: I want to do a mysqldump before the upgrade | 21:20 |
| clarkb | also going to wait for neutron meeting to end as they appear to be actively using etherpad for tracking stuff right now | 21:21 |
| *** gmurphy has joined #openstack-infra | 21:21 | |
| openstackgerrit | Michael Krotscheck proposed openstack-infra/storyboard: Updated oslo_db. https://review.openstack.org/173083 | 21:22 |
| openstackgerrit | Michael Krotscheck proposed openstack-infra/storyboard: Removed hack. https://review.openstack.org/173084 | 21:22 |
| *** mattfarina has quit IRC | 21:23 | |
| krotscheck | greghaynes: ^^ There ya go | 21:24 |
| *** dustins has quit IRC | 21:25 | |
| greghaynes | krotscheck: oh, does storyboard not use global-reqs? | 21:25 |
| krotscheck | greghaynes: Nope | 21:26 |
| krotscheck | greghaynes: We're not really openstack | 21:26 |
| krotscheck | greghaynes: Well, that's not really true. We're a bit of a special snowflake. | 21:26 |
| clarkb | its not that special, all of infra is that way | 21:26 |
| krotscheck | clarkb: shaddup. we're a special snowflake. | 21:26 |
| krotscheck | SPECIAL I SAY | 21:27 |
| clarkb | there is a difference between deploying cloud software and running developer tooling on a cloud | 21:27 |
| openstackgerrit | Khai Do proposed openstack-infra/puppet-gerrit: Add properties, md and css files to direct download list https://review.openstack.org/173087 | 21:27 |
| greghaynes | yea, I dont think its a bad thing, just a TIL | 21:27 |
| clarkb | zaro: can you make sure that the bouncy castle upgrade from gerrit 2.10 ends up on the meeting agenda tomorrow? | 21:27 |
| * krotscheck blinks. | 21:28 | |
| krotscheck | bouncy castle? | 21:28 |
| clarkb | krotscheck: yes, it is unfortunately not as awesome as the name might imply https://www.bouncycastle.org/java.html | 21:28 |
| *** Sukhdev has quit IRC | 21:28 | |
| *** fedexo has joined #openstack-infra | 21:28 | |
| krotscheck | clarkb: Huhn. i dunno, I kindof feel that a non-US-export-restricted version of crypto tools is awesome. | 21:29 |
| krotscheck | But then I'm an anarchist :D | 21:29 |
| * krotscheck isn't really an anarchst | 21:30 | |
| clarkb | krotscheck: oh sure, we have plenty of those. This particular one appeas to release backward incompatible changes that force us to upgrade out of band with our distros which is a giant headache | 21:30 |
| krotscheck | clarkb: Bring them under the openstack banner. | 21:30 |
| krotscheck | clarkb: :) | 21:30 |
| zaro | clarkb: well, it's on the actions list for me to test gerrit 2.10 upgrade. so we should hit it. | 21:30 |
| clarkb | zaro: awesome | 21:30 |
| *** isviridov_away is now known as isviridov | 21:31 | |
| hashar | fungi: not sure whaqt is the best to tag :) | 21:31 |
| *** mmedvede has joined #openstack-infra | 21:32 | |
| fungi | hashar: i think i'll probably just push to do the branch tip | 21:32 |
| hashar | fungi: but I am surely actively working on the Debian packaging :D I will post on your list once I have something worth sharing | 21:32 |
| *** gyee has quit IRC | 21:32 | |
| clarkb | cinerama: how are we doing with zanata-dev? anything I can do while I wait for reviews myself? | 21:32 |
| fungi | hashar: that's awesome--thanks!!! | 21:32 |
| hashar | fungi: and we have almost all our CI using a Zuul debian package :D | 21:32 |
| hashar | fungi: i am migrating the scheduler this week.; Then will have to catch up with 3 or 4 months of your changes | 21:32 |
| *** mrda-away is now known as mrda | 21:33 | |
| greghaynes | krotscheck: I think your oslo.db version hasnt been synced to mirrors yet | 21:33 |
| *** mjturek1 has joined #openstack-infra | 21:34 | |
| krotscheck | greghaynes: Likely | 21:34 |
| krotscheck | I'll recheck later | 21:34 |
| *** asettle-gym is now known as asettle | 21:34 | |
| *** dizquierdo has joined #openstack-infra | 21:34 | |
| *** dizquierdo has quit IRC | 21:34 | |
| *** dizquierdo has joined #openstack-infra | 21:34 | |
| fungi | krotscheck: greghaynes: weird. we only got the wheel not the sdist tarball http://pypi.dfw.openstack.org/simple/oslo.db/ | 21:35 |
| fungi | oh, that's because pypi only got the wheel https://pypi.python.org/pypi/oslo.db | 21:35 |
| fungi | dhellmann: ^ | 21:35 |
| fungi | i'm checking the upload logs now | 21:36 |
| *** frickler_ has joined #openstack-infra | 21:36 | |
| dhellmann | fungi: hrm | 21:36 |
| * dhellmann notes that pypi started showing dependencies at some point when he wasn't looking | 21:36 | |
| fungi | ick! http://logs.openstack.org/62/6235b7ed3957eb98e67f02bf761535af37d23123/release/oslo.db-pypi-both-upload/1e75289/console.html | 21:37 |
| fungi | HTTPError: 503 Server Error: backend read error | 21:37 |
| clarkb | pypi said "NO" | 21:37 |
| greghaynes | womp womp | 21:37 |
| fungi | dstufft: ^ is it safe to retry twine when that happens? | 21:37 |
| *** frickler has quit IRC | 21:38 | |
| * krotscheck feels like the canary in the coalmine :) | 21:38 | |
| *** miqui has quit IRC | 21:38 | |
| openstackgerrit | Michael Krotscheck proposed openstack-infra/storyboard: Updated oslo_log https://review.openstack.org/173093 | 21:38 |
| openstackgerrit | Michael Krotscheck proposed openstack-infra/storyboard: Updated all oslo versions. https://review.openstack.org/173094 | 21:38 |
| *** fedexo has quit IRC | 21:39 | |
| dstufft | fungi: probably | 21:41 |
| fungi | dstufft: strangely, it seems to have uploaded that file, but returned a 503 | 21:41 |
| *** Rockyg has joined #openstack-infra | 21:42 | |
| *** kgiusti has left #openstack-infra | 21:42 | |
| dstufft | fungi: 503 backend read error means "the internet between fastly and PyPI went sideways during the request" | 21:42 |
| *** mriedem is now known as mriedem_away | 21:42 | |
| fungi | wondermous | 21:42 |
| dstufft | so it could happen during the process of uploading | 21:42 |
| dstufft | or it could happen after that before fastly can read the response | 21:42 |
| dstufft | and techincally it looks like Fastly Edge -> Fastly Shield -> PyPI, so it could happen between Fastly Edge and Fastly Shield, and the request to Fastly Shield -> Fastly PyPI still happens | 21:43 |
| *** e0ne has quit IRC | 21:43 | |
| dstufft | tl;dr it depends | 21:43 |
| fungi | looks like the md5sum of whay i downloaded from pypi matches what twine tried to upload, so it's complete | 21:44 |
| *** isviridov is now known as isviridov_away | 21:44 | |
| *** cody-somerville has joined #openstack-infra | 21:45 | |
| *** gordc has quit IRC | 21:45 | |
| fungi | dhellmann: okay, i've manually rerun the upload steps for the tarball and it's showing at https://pypi.python.org/pypi/oslo.db now | 21:48 |
| fungi | should end up on our mirrors in moments | 21:49 |
| *** rtom has joined #openstack-infra | 21:49 | |
| *** rtom has quit IRC | 21:49 | |
| dhellmann | fungi: thank you. I'd like to send email about the issue. Is it safe to say something like "The upload job failed for the sdist so some installations might have failed. The missing file has been restored and should appear in our mirror shortly." | 21:49 |
| clarkb | hrm so are mordred jeblair pleia2 all afk right now? | 21:50 |
| clarkb | jhesketh: awake yet? :) | 21:50 |
| fungi | dhellmann: it may not have even been noticed by anyone, the wheel got uploaded fine so anyone using a new enough pip to grab wheels got the new release and everyone else got the previous release tarball still | 21:50 |
| dhellmann | fungi: ok, how did we notice then? | 21:50 |
| fungi | dhellmann: greghaynes and krotscheck were manually checking for a new behavior they were expecting from the latest release, and weren't seeing it, sounds like | 21:51 |
| *** fedexo has joined #openstack-infra | 21:51 | |
| fungi | anyway, it's showing up on our mirrors now too | 21:52 |
| dhellmann | fungi: ok, thanks | 21:52 |
| *** notnownikki has quit IRC | 21:53 | |
| *** dkranz has quit IRC | 21:53 | |
| greghaynes | dhellmann: fungi yep - there was a version bump on the storyboard req's so a workaround for a bug could be removed and we werent able to find the package | 21:57 |
| krotscheck | dhellmann fungi: Thanks, rechecking the patch. | 21:57 |
| krotscheck | For reference: https://review.openstack.org/#/c/173083/ | 21:58 |
| krotscheck | I love working on email. It always starts with: Oh, these other things broke in the meantime. | 21:59 |
| *** markvoelker_ has joined #openstack-infra | 22:00 | |
| *** dizquierdo has quit IRC | 22:00 | |
| *** otter768 has joined #openstack-infra | 22:02 | |
| *** crainte has quit IRC | 22:02 | |
| openstackgerrit | Doug Hellmann proposed openstack-infra/project-config: Use Oslo core teams for stable maint, too https://review.openstack.org/173075 | 22:03 |
| lifeless | clarkb: oh hai | 22:05 |
| reed | does anybody know how to diagnose a bounce in mailman? | 22:05 |
| pleia2 | clarkb: I just landed at my hotel and have a little while before the conference opening social, need me to look at something? | 22:05 |
| clarkb | lifeless: hello | 22:05 |
| *** unicell has joined #openstack-infra | 22:05 | |
| *** andreykurilin has quit IRC | 22:05 | |
| clarkb | pleia2: https://review.openstack.org/#/c/164323/ no need to approve. I want to mysqldump first, just looking for review | 22:05 |
| fungi | reed: sort of. what specifically are you looking for? | 22:05 |
| reed | every now and then there is someone whose "membership in the mailing list Openstack has been disabled due to excessive bounces" | 22:05 |
| * pleia2 has a look | 22:05 | |
| reed | fungi, how to trace back why that address was disabled? | 22:06 |
| lifeless | clarkb: hows te ovs issue? | 22:06 |
| krotscheck | I love version kickoff week. All my builds go to the top of the queue | 22:06 |
| clarkb | lifeless: I havne't looked at it today, but as far as I know nothing has changed since | 22:06 |
| *** otter768 has quit IRC | 22:06 | |
| fungi | reed: most frequent cause i've seen is subscribers with dmarc-enforcing gateways causing them to get unsubscribed because they received a message through the list from someone at a dmarc-enabled domain | 22:06 |
| clarkb | lifeless: if you point me at an ssh pubkey I can give you access to the held env | 22:06 |
| *** tonytan4ever has quit IRC | 22:07 | |
| lifeless | clarkb: the one on gerrit is fine | 22:07 |
| lifeless | clarkb: or /query | 22:07 |
| reed | fungi, but there is no way to find out after the fact? | 22:07 |
| reed | i see in bounce.log there is only the count, not the reason | 22:07 |
| fungi | reed: so the list tries to deliver the message to them and their mailserver refuses it because the dmarc signature is invalid, and after something like 5 times of that it unsubscribes them | 22:08 |
| fungi | reed: the actual rejections are in the exim logs i think. have a specific subscriber you need checked? | 22:08 |
| reed | fungi, I'll check the exim logs, I want to learn how to do this | 22:08 |
| reed | fungi, it looks like an aggressive symantec filter | 22:10 |
| reed | 553-Message filtered. Refer to the Troubleshooting page at\n553-http://www.symanteccloud.com/troubleshooting for more\n553 information. | 22:10 |
| fungi | reed: quite possible | 22:10 |
| *** mwagner_lap has quit IRC | 22:10 | |
| *** soren has quit IRC | 22:11 | |
| fungi | the user should suggest their mailserver administrators whitelist e-mail coming from our lists.openstack.org server | 22:12 |
| reed | yep, thanks | 22:13 |
| lifeless | clarkb: should I ask q's here or query ? | 22:13 |
| clarkb | lifeless here is fine | 22:13 |
| *** adalbas has quit IRC | 22:14 | |
| clarkb | lifeless: as long as we don't interrupt too much other conversation but its been pretty quiet | 22:14 |
| lifeless | clarkb: ok, so this is a nova-network config ? | 22:14 |
| clarkb | lifeless: correct | 22:14 |
| cinerama | clarkb: hey, i was having lunch | 22:14 |
| lifeless | br_pub is the shared virtual L2? | 22:15 |
| clarkb | lifeless: one of them yes. br_flat is another | 22:15 |
| lifeless | why are there two ? | 22:15 |
| clarkb | lifeless: br_pub is the L2 intended to carry the tempest to floating IP traffic though and is the one that doesn't work | 22:15 |
| clarkb | lifeless: because nova ebtables breaks if you carry flat traffic and public traffic over the same l2 | 22:16 |
| lifeless | ok | 22:16 |
| lifeless | so flat is internal l2 | 22:16 |
| clarkb | yup | 22:16 |
| lifeless | pub is external l2 | 22:16 |
| clarkb | yup | 22:16 |
| lifeless | and pub is broke | 22:16 |
| clarkb | yes, if you try to ping 172.24.4.2 from 172.24.4.1 the arp who has will hit br_pub then ovs will drop it | 22:16 |
| clarkb | it hits br_pub on 172.24.4.1, no who has crosses gre | 22:17 |
| cinerama | clarkb: so assuming the last merge got applied to translate-dev...would be good to see the /var/log/wildfly/console.log & might need my account tweaked in the database | 22:17 |
| *** asettle has quit IRC | 22:19 | |
| *** amitgandhinz has quit IRC | 22:19 | |
| lifeless | clarkb: there's no access port | 22:21 |
| clarkb | cinerama: I still see the from email exception, I probably have to restart the service? | 22:21 |
| lifeless | clarkb: is that deliberate? | 22:21 |
| clarkb | lifeless: semi, I did that because thats how neutron wants br-ex defined. But when tested with a veth pair one end attached to br_pub it fails the same way | 22:21 |
| *** dboik has quit IRC | 22:21 | |
| cinerama | clarkb: huh. wildfly should've restarted with the new config if puppet got run | 22:21 |
| clarkb | lifeless: at least it appeared to via tempest | 22:21 |
| clarkb | cinerama: let me double check how long the process has been running | 22:22 |
| lifeless | clarkb: well, neutron must not touch these ports | 22:22 |
| lifeless | clarkb: and bridges - br-ex is going to have to be layered as a port on top of br_pub | 22:22 |
| clarkb | cinerama: appears to be running since april 10th | 22:22 |
| *** dboik has joined #openstack-infra | 22:22 | |
| clarkb | lifeless: not quite, instead of calling it br_pub we just make an L2 called br-ex | 22:22 |
| lifeless | clarkb: otherwise neutron will mess things up, vs fix it :) | 22:22 |
| clarkb | lifeless: then neutron uses that and is mostly happy | 22:22 |
| cinerama | clarkb: hmm sounds like puppet hasn't run there yet...do we need to run it manually there? | 22:23 |
| *** MaxV has joined #openstack-infra | 22:23 | |
| clarkb | lifeless: I also tried connecting br-ex to our existing brctl made linux bridge using gre and that did not work but I didn't spend much time debugging that | 22:23 |
| clarkb | cinerama: oh I bet I know why, absible isn't likely able to ssh in | 22:23 |
| lifeless | clarkb: depends on what you're doing with Neutron I suspect. VLAN testing will need the CI infra to bring up the L2 fully. Still - you've tested, I shall not argue :) | 22:24 |
| lifeless | clarkb: can I fiddle here with a free hand? | 22:24 |
| clarkb | pleia2: cinerama should I make it so ansible can puppet that node? I will run puppet manually right now | 22:24 |
| clarkb | lifeless: yup, I can always rebuild a new pair of nodes in that state | 22:24 |
| clarkb | lifeless: so feel free | 22:24 |
| cinerama | clarkb: i don't see why not, i would defer to pleia2 here but i think she is busy | 22:24 |
| pleia2 | clarkb, cinerama wfm | 22:25 |
| cinerama | but not *that* busy :) | 22:25 |
| clarkb | cinerama: pleia2 puppet run by hand and wildfly restarted | 22:26 |
| clarkb | I do not see exceptions now | 22:27 |
| clarkb | cinerama: do you need to test something to determine if db needs updating? | 22:27 |
| cinerama | clarkb: i'll try logging in. it'll give *me* a vague error message but there'll be something in the logs | 22:27 |
| clarkb | cinerama: kk | 22:27 |
| clarkb | fungi: are you around? | 22:27 |
| clarkb | fungi: I am going to make an etherpad db dump now, then approve the upgrade change | 22:28 |
| *** Rockyg has quit IRC | 22:29 | |
| clarkb | fungi: I know its getting late for you, but hoping to have at least one core/root around before I upgrade | 22:29 |
| *** annegentle has quit IRC | 22:29 | |
| fungi | clarkb: totally here, just taking out the trash | 22:30 |
| *** cschwede has quit IRC | 22:30 | |
| cinerama | clarkb: i'm going to need a console log on that guy. we were going so well... | 22:31 |
| cinerama | clarkb: /var/log/wildfly/console.log | 22:31 |
| clarkb | cinerama: lots of could not prepare statement db errors | 22:32 |
| *** tiswanso has joined #openstack-infra | 22:32 | |
| *** baoli_ has quit IRC | 22:32 | |
| *** cschwede has joined #openstack-infra | 22:33 | |
| cinerama | clarkb: i haven't seen that before. i might need a bit more detail if possible. | 22:33 |
| lifeless | clarkb: ok so I know whats wrong | 22:33 |
| lifeless | clarkb: you need to have just one gre l2 bridge | 22:34 |
| clarkb | fungi: ok backup taken and is in /root/clarkb-db-dumps | 22:34 |
| clarkb | fungi: I am approving the change now | 22:34 |
| fungi | clarkb: thanks, noted | 22:34 |
| lifeless | clarkb: and then wrap one or both of _pub and _flat with VLAN tags. | 22:34 |
| clarkb | lifeless: thats silly | 22:34 |
| lifeless | clarkb: no its not | 22:35 |
| clarkb | lifeless: they are two "physically" separated bridged | 22:35 |
| lifeless | clarkb: not right now they aren't | 22:35 |
| clarkb | lifeless: you don't need vlan tagging in that case | 22:35 |
| clarkb | lifeless: really? what is bridging them? | 22:35 |
| lifeless | clarkb: you have two gre tunnels on the same IP address at both end | 22:35 |
| clarkb | oh hrm does ovs not let us distinguish those as different tunnels like the ip made gre interfaces? | 22:36 |
| *** gyee has joined #openstack-infra | 22:36 | |
| lifeless | clarkb: write down the expected packet headers from each tunnel - src ip, dst ip, ip proto | 22:36 |
| lifeless | clarkb: then tell me which bridge its for | 22:36 |
| *** rm_work is now known as rm_work|away | 22:37 | |
| clarkb | lifeless: https://git.openstack.org/cgit/openstack-infra/devstack-gate/tree/functions.sh#n825 note the key | 22:37 |
| clarkb | lifeless: this should be doable, but ovs either can't do it or has incomplete configuration | 22:37 |
| *** tiswanso has quit IRC | 22:38 | |
| lifeless | GRE tunnel keys are what ovs maps traffic tags to | 22:38 |
| *** marun has quit IRC | 22:38 | |
| clarkb | lifeless: it would be amazing if ovs documented that anywhere | 22:38 |
| fungi | as opposed to the interface ids | 22:39 |
| fungi | surprising, but i can see reasons for that choice | 22:40 |
| clarkb | also it would be nice if it let arp through instead of dropping it wihtout a reason | 22:40 |
| lifeless | this is going off of memory, but i'm fairly sure.. can test by adding a vlan access port to the working bridge | 22:41 |
| lifeless | clarkb: I think its blocked everything off from some sanity check somewhere | 22:41 |
| lifeless | give me a minute to experiment | 22:41 |
| clarkb | lifeless: psuedo spanning tree? | 22:41 |
| *** MaxV has quit IRC | 22:41 | |
| lifeless | huh, no | 22:42 |
| clarkb | cinerama: working on getting a paste that isn't too crazy for you | 22:42 |
| clarkb | cinerama: I think i have found where the interesting stuff starts | 22:43 |
| cinerama | clarkb: cool, thanks. | 22:43 |
| openstackgerrit | Merged openstack-infra/system-config: Ensure latest on etherpad vcsrepo https://review.openstack.org/164323 | 22:43 |
| fungi | so basically one gre tunnel with two (or later more if we want) 802.1q encapsulated vlans, one for flat and one for pub? | 22:44 |
| *** zz_dimtruck is now known as dimtruck | 22:44 | |
| clarkb | fungi: well lifless just said "huh, no" so maybe not :) | 22:45 |
| clarkb | fungi: but if that is indeed the issue yes, we would set it up that way | 22:45 |
| fungi | huh, ok ;) | 22:45 |
| clarkb | cinerama: its about 1k lines | 22:45 |
| lifeless | yeah, default port is trunk | 22:45 |
| lifeless | thought so | 22:45 |
| fungi | yeah, it'll cost us a handful of additional bytes out of the mtu, but we're there regardless with gre | 22:45 |
| cinerama | clarkb: bring it ooooon. there shouldn't be anything too heinous in there | 22:45 |
| lifeless | btw ovs-vswitchd.conf.db is where a lot of stuff is documented | 22:45 |
| cinerama | clarkb: it's java. they love their verbose errors | 22:46 |
| lifeless | clarkb: from readingthat in_key and out_key can be used to control the tunnel key analogous to the ip commands | 22:47 |
| clarkb | cinerama: http://paste.openstack.org/show/203776/ I actually trimmed it down a bit more, realized you likely don't need to see all the startup stuff | 22:47 |
| lifeless | clarkb: but, I think it matches the reality better to think of the tunnel as a trunk port in a dc | 22:47 |
| clarkb | lifeless: I don't | 22:47 |
| clarkb | lifeless: its a point to point tunnel | 22:47 |
| clarkb | lifeless: saying it matches reality better to carry tagged traffic isn't more or less correct that using multiple point to points | 22:48 |
| clarkb | lifeless: just like you can connect two devices in a DC on different ports without tagging or you can use a single port with tagging | 22:48 |
| lifeless | clarkb: there is no physical separation, so if one is used in tests with VLANs, that overlap a manually chosen key, it *will* conflict. | 22:48 |
| lifeless | clarkb: this is precisely the behaviour of a trunk port in a DC | 22:48 |
| clarkb | both are equally valid realities | 22:48 |
| fungi | trunk ports rarely have ip addresses on them | 22:49 |
| cinerama | clarkb: Caused by: com.mysql.jdbc.exceptions.jdbc4.CommunicationsException: Communications link failure | 22:49 |
| fungi | on their endpoints i mean | 22:49 |
| lifeless | if there were multiple points, I would agree with you | 22:49 |
| lifeless | get another public IP for the VM, and what you are describing would make sense | 22:49 |
| lifeless | clarkb: I give up | 22:49 |
| lifeless | clarkb: I'll check this works, and hand back to you | 22:49 |
| clarkb | lifeless: I just don't want to make decisions based on arbitrary things that don't matter | 22:50 |
| clarkb | lifeless: both are correct. We can use either. Lets not say one is better because data center | 22:50 |
| fungi | on the other hand "this works, the other doesn't" is far from arbitrary. but yes the analogies aren't critical to the solution | 22:50 |
| lifeless | they aren't both correct, we can't use either unless one of them is not being used as a raw L2 | 22:51 |
| cinerama | clarkb: wondering if the trove is ok | 22:51 |
| clarkb | lifeless: I thought you said we could either use vlan tagging our in_key and out_key? | 22:51 |
| lifeless | from the docs | 22:51 |
| clarkb | s/our/or/ | 22:51 |
| lifeless | Each tunnel must be uniquely identified by the combination of type, options:remote_ip, | 22:51 |
| lifeless | options:local_ip, and options:in_key. | 22:51 |
| *** mtanino has quit IRC | 22:51 | |
| *** coolsvap has quit IRC | 22:51 | |
| lifeless | clarkb: I said *as long as* we're not putting tagged traffic on *either* of them | 22:51 |
| clarkb | cinerama: I can give it a go in a moment | 22:51 |
| fungi | this reminds me more of how some switch-router platforms handle l2 vpn id keys directly in their bridge tables | 22:51 |
| fungi | i'm glaring at you, screenos | 22:52 |
| lifeless | clarkb: there are both neutron and nova-net configurations that will do that, and that will break with one config in mysterious ways, but not in the other. | 22:52 |
| cinerama | clarkb: thanks, just when you're free | 22:52 |
| clarkb | lifeless: I guess I still don't follow, why can't we use in_key on both of them with different keys? | 22:53 |
| lifeless | IFF all the ports are access ports, VLANs prohibited, because the underlying substrate is acting like a trunk port | 22:54 |
| clarkb | lifeless: but the substrates are independent | 22:54 |
| lifeless | they are not | 22:54 |
| clarkb | lifeless: br_pub is one switch with its own flows and br_flat is another | 22:54 |
| lifeless | they are not the substrates | 22:54 |
| lifeless | GRE is the subtrate | 22:54 |
| openstackgerrit | Clint 'SpamapS' Byrum proposed openstack-infra/shade: Add test for invalidation after delete https://review.openstack.org/172600 | 22:55 |
| openstackgerrit | Clint 'SpamapS' Byrum proposed openstack-infra/shade: Deprecate use of cache in list_volumes https://review.openstack.org/172599 | 22:55 |
| openstackgerrit | Clint 'SpamapS' Byrum proposed openstack-infra/shade: Invalidate volume list cache when creating https://review.openstack.org/172595 | 22:55 |
| openstackgerrit | Clint 'SpamapS' Byrum proposed openstack-infra/shade: Make cache key generator ignore cache argument https://review.openstack.org/173109 | 22:55 |
| clarkb | lifeless: but both would be uniquely identified by different keys | 22:55 |
| lifeless | as long as ovs knows everything is access ports | 22:56 |
| lifeless | but you're fighting the way ovs is designed | 22:56 |
| clarkb | I am? | 22:56 |
| clarkb | all I wanted was two logical l2s completely separated from each other in a manner that brctl supports | 22:56 |
| *** dprince has quit IRC | 22:56 | |
| reed | can someone please review these two? a little more documentation would be good https://review.openstack.org/#/c/104951/ https://review.openstack.org/#/c/138195/ | 22:56 |
| clarkb | if ovs wants to be different it should document that so that its poor users can figure these different design decisions | 22:57 |
| jhesketh | Morning | 22:57 |
| clarkb | if I have to use vlan tagging thats fine, but none of the documentation/man pages/blog posts I could find mentioned that | 22:57 |
| lifeless | clarkb: it is documented - see the man page I referenced above | 22:57 |
| lifeless | clarkb: or I may be wrong, it may be that all the things I've looked at before were misunderstood | 22:58 |
| lifeless | clarkb: *anyway*, using in_key and out_key should work until VLANs are being tested, and then it may or may not work | 22:59 |
| lifeless | right now it can't work because there are two identical tunnel definitions and that is documented as being illegal | 22:59 |
| clarkb | lifeless: reading the man page you pointed out does not explain anything about access ports vs other ports | 22:59 |
| lifeless | line 1775 | 23:00 |
| lifeless | and around line 768 | 23:00 |
| lifeless | the key thing needed to match the ip commands is to set in_key and out_key | 23:00 |
| lifeless | thatwill make the tunnels unique | 23:01 |
| openstackgerrit | Clint 'SpamapS' Byrum proposed openstack-infra/shade: MonkeyPatch time.sleep in unit tests to avoid wait https://review.openstack.org/173110 | 23:01 |
| lifeless | in_key:flow is what does flow -> key mapping | 23:01 |
| lifeless | clarkb: ^ that is where I think we've been talking at cross purpose | 23:01 |
| *** bknudson has quit IRC | 23:02 | |
| *** hashar has quit IRC | 23:02 | |
| lifeless | clarkb: so - to fix your patchset, you need to set in_key and out_key on both those bridges to unique values | 23:02 |
| clarkb | yes tahts exactly what I said and you said I needed to do vlan tagging | 23:02 |
| lifeless | clarkb: yes, I was used to flow based keys | 23:02 |
| lifeless | clarkb: I didn't realise that that was optional | 23:03 |
| *** dims_ has joined #openstack-infra | 23:03 | |
| *** cdent has quit IRC | 23:03 | |
| *** ftersin has quit IRC | 23:03 | |
| lifeless | regardless, the issue is clearly diagnosed - yes? | 23:03 |
| clarkb | I think so, about to push a patchset up to confirm | 23:03 |
| lifeless | the (gre, remote_ip, local_ip, in_key) tuple must be unique | 23:03 |
| *** dims has quit IRC | 23:04 | |
| *** mtanino has joined #openstack-infra | 23:04 | |
| lifeless | and in_key on one must match out_key on the other | 23:04 |
| clarkb | up and you can use key as a shortcut to make them both match | 23:04 |
| lifeless | yup | 23:04 |
| fungi | this almost smells like juniper had a hand in the openvswitch specification | 23:05 |
| lifeless | clarkb: so - tl;dr: problem diagnosed, solution proposed, and I'm sorry I was a grumpy f*ing bear | 23:05 |
| *** shashankhegde has quit IRC | 23:05 | |
| clarkb | lifeless: thank you for looking. | 23:06 |
| openstackgerrit | Clint 'SpamapS' Byrum proposed openstack-infra/shade: MonkeyPatch time.sleep in unit tests to avoid wait https://review.openstack.org/173110 | 23:06 |
| clarkb | I just need to figure out how to supply multiple options now | 23:07 |
| SpamapS | Shrews: Ran 23 tests in 0.113s (+0.006s) | 23:07 |
| * SpamapS just can't stand tests that time.sleep ;) | 23:07 | |
| *** dimtruck is now known as zz_dimtruck | 23:07 | |
| *** shashankhegde has joined #openstack-infra | 23:07 | |
| *** Krinkle is now known as Krinkle|detached | 23:08 | |
| *** cody-somerville has quit IRC | 23:09 | |
| *** dannywilson has joined #openstack-infra | 23:09 | |
| lifeless | clarkb: I'm all logged out of those machines now | 23:10 |
| clarkb | lifeless: kk, I will clean them up as soon as I get results back from the patch I am about to push | 23:10 |
| clarkb | lifeless: any idea how to pass multiple options when using ovs-vsctl? ovs-vsctl man page does not seem to cover this (or I skimmed too fast) | 23:11 |
| clarkb | looks like its options:key=val options:key=val repeated | 23:12 |
| clarkb | thank you pica8 | 23:13 |
| lifeless | clarkb: what options are you setting | 23:13 |
| lifeless | yah | 23:13 |
| clarkb | lifeless: remote_ip and key | 23:13 |
| *** dannywilson has quit IRC | 23:13 | |
| *** rbradfor_ has quit IRC | 23:13 | |
| clarkb | so options:remote_ip=x.x.x.x options:key=X seems to do the trick | 23:13 |
| *** chlong has joined #openstack-infra | 23:15 | |
| lifeless | cool | 23:15 |
| lifeless | might be an idea to also set local_ip | 23:15 |
| lifeless | to avoid sending the traffic out the wrong port in the event of eth1 going down or some such | 23:15 |
| openstackgerrit | Clark Boylan proposed openstack-infra/devstack-gate: DO NOT MERGE testing overlay network options https://review.openstack.org/171714 | 23:15 |
| clarkb | oh, well we can try that if ^ still has trouble | 23:15 |
| lifeless | oh, Im' suggesting it as belts-n-bracers, since it won't be needed on any happy-path occasion | 23:16 |
| clarkb | gotcha | 23:16 |
| lifeless | hmm, pbr tests failure on 2.7 | 23:17 |
| lifeless | also - hope everyone saw the shiny shiny news that pip now autobuilds wheels :) | 23:17 |
| clarkb | fungi: looks like puppet updated git but didn't restart the service for etherpad | 23:18 |
| clarkb | fungi: http://puppetboard.openstack.org/report/etherpad.openstack.org/1235f7bf8e3ce87f7fe90e48e1f91932a2305410 I am going to restart it now | 23:18 |
| fungi | k | 23:18 |
| clarkb | ok should be up now | 23:20 |
| clarkb | please test | 23:20 |
| fungi | doing | 23:20 |
| clarkb | ok first thing I notice is I need a hard refresh otherwise not all needed js ends up in my browser | 23:21 |
| clarkb | so yay caches | 23:21 |
| clarkb | https://etherpad.openstack.org/p/clarkb-test can be our test pad | 23:21 |
| fungi | lgtm so far. i didn't have any pads open/cached | 23:21 |
| *** jamielennox|away is now known as jamielennox | 23:22 | |
| *** dboik_ has joined #openstack-infra | 23:22 | |
| *** dboik_ has quit IRC | 23:22 | |
| fungi | had just restarted my browser because of fat fingers (ctrl-w to delete the last word is dangerously close to ctrl-q on a qwerty board) | 23:22 |
| clarkb | ok I will send a note to the dev list now saying we updated it and if you see errors please hard refresh to invalidate any cached javascript | 23:22 |
| *** dboik_ has joined #openstack-infra | 23:22 | |
| mtreinish | my red looks brighter than usual to me | 23:23 |
| mtreinish | but it's probably all in my head | 23:23 |
| *** rlandy has quit IRC | 23:23 | |
| clarkb | cinerama: ok I am almost back to where I can help with translate dev | 23:24 |
| clarkb | cinerama: give me a few more minutes (haven't forgotten about you) | 23:24 |
| *** rbradfor_ has joined #openstack-infra | 23:24 | |
| *** rbradfor_ has quit IRC | 23:24 | |
| clarkb | fungi: have a list of CVEs I can attribute the update to? | 23:24 |
| fungi | clarkb: so far just one, because mitre is being slow | 23:24 |
| clarkb | I will take that one :) | 23:25 |
| fungi | clarkb: CVE-2015-3297 | 23:25 |
| *** dboik has quit IRC | 23:25 | |
| Shrews | SpamapS: awesome. good catch | 23:27 |
| *** cody-somerville has joined #openstack-infra | 23:27 | |
| clarkb | fungi: mail sent to the -dev list thanks for your help | 23:29 |
| fungi | clarkb: thank you! | 23:29 |
| clarkb | hrm I think I sent it, don't see it yet. That'll learn me to use a silly mua | 23:30 |
| clarkb | http://lists.openstack.org/pipermail/openstack-dev/2015-April/061393.html there it is | 23:30 |
| mtreinish | mordred: man that migration we wrote is really slow. I've been running it on a much smaller dataset in my test db on hp cloud and it's been going for > 3hrs already | 23:30 |
| mtreinish | mordred: although now that I think about it, it's probably hp cloud just being slow | 23:30 |
| *** amitgandhinz has joined #openstack-infra | 23:30 | |
| *** wenlock1 has quit IRC | 23:32 | |
| *** wenlock has quit IRC | 23:33 | |
| clarkb | cinerama: I am able to connect to the db using the zanata user and supplied passwd | 23:34 |
| clarkb | cinerama: maybe the url connection string is wrong? | 23:34 |
| *** amitgandhinz has quit IRC | 23:35 | |
| *** mwagner_lap has joined #openstack-infra | 23:35 | |
| cinerama | clarkb: that's interesting. can you have a look at /opt/wildfly/standalone/configuration/standalone.xml? | 23:35 |
| clarkb | cinerama: thats where I took the connection info from | 23:35 |
| clarkb | cinerama: but I used mysql client not jdbc to connect so had to map things | 23:36 |
| *** marun has joined #openstack-infra | 23:36 | |
| cinerama | clarkb: right, the username and password are in the <security> section under <subsystem xmlns="urn:jboss:domain:datasources:2.0"> | 23:37 |
| clarkb | yup | 23:37 |
| clarkb | cinerama: I am wondering if the connection url is wrong though | 23:37 |
| clarkb | cinerama: its jdbc:mysql://host:3306/zanata?characterEncoding=UTF-8 | 23:37 |
| cinerama | clarkb: is "host" a bowdlerization of the actual host? | 23:38 |
| clarkb | cinerama: yes | 23:38 |
| clarkb | cinerama: the hostname is also correct | 23:39 |
| clarkb | I used it verbatim from the config to connect with mysql client | 23:39 |
| *** SergeyLukjanov has quit IRC | 23:39 | |
| cinerama | clarkb that should be OK. we haven't changed that bit of the template in many revs. | 23:39 |
| *** nbernard has quit IRC | 23:39 | |
| *** asettle has joined #openstack-infra | 23:40 | |
| asettle | Good morning again! | 23:40 |
| *** SergeyLukjanov has joined #openstack-infra | 23:40 | |
| *** asettle has left #openstack-infra | 23:41 | |
| *** Swami has quit IRC | 23:41 | |
| cinerama | clarkb: and now i'm in. weird | 23:41 |
| clarkb | cinerama: you're into the db or into zanata? | 23:41 |
| clarkb | looks like zanata (looking at logs) | 23:42 |
| cinerama | clarkb: yes, i've logged in..the only downside is i typoed my username so i don't have admin powerz | 23:42 |
| clarkb | well I am tailing the log now will let you know if I see db issues again | 23:42 |
| *** r-daneel has quit IRC | 23:43 | |
| clarkb | there the go | 23:45 |
| clarkb | I wonder if this is realted to the issue review had with trove | 23:45 |
| clarkb | the db times out too fast | 23:45 |
| clarkb | pleia2: did you apply the config to the db and restart the db? | 23:45 |
| clarkb | cinerama: I will poke at trove settings in a bit | 23:46 |
| *** rlucio_ has quit IRC | 23:46 | |
| cinerama | clarkb: it's also not lvoing my login | 23:47 |
| *** marun has quit IRC | 23:47 | |
| *** markvoelker has joined #openstack-infra | 23:49 | |
| *** zz_dimtruck is now known as dimtruck | 23:49 | |
| *** markvoelker_ has quit IRC | 23:51 | |
| *** weshay has quit IRC | 23:52 | |
| *** markvoelker_ has joined #openstack-infra | 23:53 | |
| *** markvoelker_ has quit IRC | 23:54 | |
| openstackgerrit | Matthew Treinish proposed openstack-infra/subunit2sql: Add db api methods to delete old runs and test_runs https://review.openstack.org/172190 | 23:55 |
| openstackgerrit | Matthew Treinish proposed openstack-infra/subunit2sql: Add a cli subcommand for deleting old rows https://review.openstack.org/172200 | 23:55 |
| openstackgerrit | Jamie Lennox proposed openstack-infra/project-config: Add django-openstack-auth-kerberos project https://review.openstack.org/172803 | 23:55 |
| *** dboik_ has quit IRC | 23:55 | |
| *** dboik has joined #openstack-infra | 23:56 | |
| *** markvoelker has quit IRC | 23:56 | |
| *** salv-orl_ has quit IRC | 23:57 | |
| clarkb | cinerama: it did not have the sanity config applied, I applied it an am restarting the db now | 23:58 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!