Friday, 2014-04-18

*** norman has joined #openstack-oslo00:05
*** ilives has joined #openstack-oslo00:13
*** ilives has quit IRC00:17
*** ilives has joined #openstack-oslo00:17
*** markmcclain has quit IRC00:41
*** arnaud has quit IRC00:46
*** arnaud has joined #openstack-oslo00:55
*** norman has quit IRC00:59
*** dims has quit IRC01:08
*** dims has joined #openstack-oslo01:20
*** mriedem has joined #openstack-oslo01:28
openstackgerritJoshua Harlow proposed a change to openstack/taskflow: Start working on switch (conditional) - WIP  https://review.openstack.org/8741701:30
*** norman has joined #openstack-oslo01:37
openstackgerritJoshua Harlow proposed a change to openstack/taskflow: Start working on switch (conditional) - WIP  https://review.openstack.org/8741701:47
*** harlowja is now known as harlowja_away02:49
dhellmannSergeyLukjanov: nevermind, I shouldn't mix beer, meetups, and IRC all at the same time -- it leaves me more confused than usual02:57
dimsdhellmann, good night we can talk about it tomorrow. i've opened a bug https://bugs.launchpad.net/nova/+bug/130929703:01
dhellmanndims: ok, thanks03:01
dhellmanndims: what do you think about process changes? What broke here?03:03
dimsdhellmann, there was a flag in nova.conf which is no longer honored03:04
dhellmannwell, yeah, I meant what should we have done differently. I don't think we're going to get very far holding up every incubator change for all of the projects to accept merges from us.03:05
dhellmann^for^waiting for03:05
dimsdhellmann, there was a -2 and a request to re-instate from nova folks. which is easier to follow up than a languishing review03:06
dimsrackspace specific eventlet flag is yuck. but folks should have debated that when it went in03:07
dhellmanntrue03:07
dhellmannyeah, we discussed that module at the summit and the dependency on eventlet was called out as something we wanted to remove to make the library more generally usable03:09
dhellmannI don't know the history of how it was added in the first place, though.03:09
dimsdhellmann, agree that it was the right thing to do and agree that it should not be part of oslo.db. but this was still common code in oslo-incubator that is synced and a -2 from a consuming project should have been dealt with before a release happened03:13
dimslet's talk more tomorrow. turning in now. good night03:13
dhellmanndims: true -- ok, ttyl03:13
*** rpodolyaka1 has joined #openstack-oslo04:02
*** mriedem has left #openstack-oslo04:03
*** zhiyan_ is now known as zhiyan04:04
SergeyLukjanovdhellmann, ok :)04:19
*** zhiyan is now known as zhiyan_04:41
*** arnaud has quit IRC04:55
openstackgerritJoshua Harlow proposed a change to openstack/taskflow: Allow for endpoints to be created at runtime  https://review.openstack.org/8816505:35
*** zhiyan_ is now known as zhiyan05:42
*** rpodolyaka1 has quit IRC05:49
*** rpodolyaka1 has joined #openstack-oslo06:09
openstackgerritJoshua Harlow proposed a change to openstack/taskflow: Be better at failure tolerance  https://review.openstack.org/8851006:12
openstackgerritJoshua Harlow proposed a change to openstack/taskflow: Be better at failure tolerance  https://review.openstack.org/8851006:16
*** rpodolyaka1 has quit IRC06:21
*** viktors has joined #openstack-oslo06:54
openstackgerritJoshua Harlow proposed a change to openstack/taskflow: Be better at failure tolerance  https://review.openstack.org/8851006:56
*** flaper87|afk is now known as flaper8707:01
*** rpodolyaka1 has joined #openstack-oslo07:21
*** rpodolyaka1 has quit IRC07:25
*** morganfainberg is now known as morganfainberg_Z07:32
*** nacim has joined #openstack-oslo07:55
*** I159 has joined #openstack-oslo08:00
*** nacim has quit IRC08:02
*** nacim has joined #openstack-oslo08:05
*** norman has quit IRC08:11
*** rpodolyaka1 has joined #openstack-oslo08:22
*** ihrachyshka has joined #openstack-oslo08:25
*** rpodolyaka1 has quit IRC08:28
openstackgerritAndreas Jaeger proposed a change to openstack/oslo-incubator: Improve help strings  https://review.openstack.org/7199209:00
*** ihrachyshka has quit IRC09:11
openstackgerritAndreas Jaeger proposed a change to openstack/oslo-incubator: Improve help strings  https://review.openstack.org/7199209:14
*** ihrachyshka has joined #openstack-oslo09:55
*** I159 has quit IRC10:00
*** 18VAAIBHG has joined #openstack-oslo10:29
*** 18VAAIBHG is now known as I15910:31
*** I159 has left #openstack-oslo10:31
openstackgerritVictor Sergeyev proposed a change to openstack/oslo-incubator: Mark db code as graduating  https://review.openstack.org/8855111:24
*** rpodolyaka has quit IRC11:28
*** rpodolyaka has joined #openstack-oslo11:28
*** gcb_ has joined #openstack-oslo12:04
*** norman has joined #openstack-oslo12:04
*** gcb_ has quit IRC12:25
openstackgerritThomas Bechtold proposed a change to openstack/oslo-incubator: Improve error handling for conf sample generation  https://review.openstack.org/8856412:34
*** mriedem has joined #openstack-oslo12:35
*** flaper87 is now known as flaper87|afk12:44
openstackgerritJulia Varlamova proposed a change to openstack/oslo-incubator: Add index modifying methods  https://review.openstack.org/6568912:49
openstackgerritVictor Sergeyev proposed a change to openstack/oslo-incubator: Remove common context usage from db model_query()  https://review.openstack.org/7965312:55
*** jecarey has quit IRC12:58
*** dims has quit IRC13:09
*** dims has joined #openstack-oslo13:21
viktorsdhellmann: around?13:30
dhellmannhi, viktors13:30
viktorsdhellmann: hell13:30
viktors*hello13:30
dhellmannheh13:30
viktors:)13:30
dhellmannwe should put together a plan for addressing the nova config issue13:31
dhellmannleaving it in nova for now seems fine, but maybe a version of that class can go into oslo.concurrency13:31
dhellmannthoughts?13:31
viktorssorry, what do you mean by oslo.concurrency ?13:32
dhellmannoh, the library that will hold some of the other oslo code related to running things in parallel13:32
dhellmannhttps://wiki.openstack.org/wiki/Oslo/GraduationStatus#oslo.concurrency13:33
viktorsoh, I missed this13:33
dhellmannwell, it wasn't clear that the db option was a candidate, since I thought only nova was using it and it was experimental13:33
dhellmannif it's critical to one of the big deployers, though, we should probably accommodate it13:34
viktorsyes, I agree, that something like oslo.concurrency should be more suitable place for eventlet.tpool13:35
dhellmannviktors: this also points out that one thing we will need to track as we move code around and make other changes is blocked patches in the other project13:35
dhellmannok, good13:35
viktorsdhellmann: yes, but if you mean Nova - we made a patch for it at December, but it was not merged, unfortunately. We should make a such  sync from oslo-incubator 'atomic'13:38
dhellmannviktors: yes, that's a good idea, too13:38
dhellmannI saw rpodolyaka's note about that on the mailing list, and I agree13:38
*** bknudson has joined #openstack-oslo13:38
viktorsdhellmann: ok, so we will consider this in the future. Maybe we need to note this in "hot-to-sync" document also?13:40
dhellmannyes, that's a good idea, too13:41
* dhellmann likes documentation13:41
viktorsok, I'll edit this page - https://wiki.openstack.org/wiki/Oslo#Syncing_Code_from_Incubator13:43
viktorsdhellmann: also I'm going to send an e-mail with review request for oslo.db repository. Should I mention something specific in this letter?13:46
*** jecarey has joined #openstack-oslo13:50
dhellmannviktors: if you had to make any changes in the repository to get the tests to pass, you could point out those changesets (since they won't be reviewed in gerrit)13:52
dhellmannviktors: other than that, maybe include a link to the blueprint so it's easy to find?13:52
viktorsdhellmann: the main changes in the tests were: 1) add common modules; 2) add entry points 3) fix  imports (openstack.common -> oslo.db.openstack.common)13:56
viktorsSo I should ask to review this commit?13:57
viktorsdhellmann: by the way, https://github.com/malor/oslo.db/commit/276f7570d7af4a7a62d0e1ffb4edf904cfbf060013:57
*** mriedem1 has joined #openstack-oslo14:03
*** mriedem has quit IRC14:04
openstackgerritIlya Pekelny proposed a change to openstack/oslo-incubator: Fix the test using in-file SQLite database  https://review.openstack.org/8759714:06
openstackgerritIlya Pekelny proposed a change to openstack/oslo-incubator: Opportunistic migration tests  https://review.openstack.org/7510714:06
openstackgerritIlya Pekelny proposed a change to openstack/oslo-incubator: Opportunistic migration tests  https://review.openstack.org/7510714:18
*** markmcclain has joined #openstack-oslo14:20
*** ilives has quit IRC14:40
dhellmannviktors: yes, including links to those commits would help with the review14:42
dhellmannviktors: you can also ask people to give their +1 on https://review.openstack.org/#/c/78955/ when they feel the repository is ready to be imported14:43
*** norman has quit IRC14:47
viktorsdhellmann: well.. The mail is already sent =)15:01
dhellmannviktors: ok, I'll follow up to your note then, no problem15:02
viktorsdhellmann: ok, I'll go on weekend then. Thanks for your help!15:03
dimshave a good weekend viktors15:03
*** creiht has joined #openstack-oslo15:03
dhellmannviktors: enjoy your weekend, and thank you!15:03
creihtdhellmann: howdy15:04
dhellmannhi, creiht15:04
*** mriedem has joined #openstack-oslo15:04
viktorsdims, dhellmann - have a good weekend too!15:04
dhellmanncreiht: thanks for volunteering to work with us this cycle15:04
creihtno problem15:04
creihtI'll try to be too much trouble ;)15:05
creihttry not to that is15:05
creihtheh15:05
dhellmannhaha15:05
dhellmannwell, we'll be relying on you to raise issues, so don't be afraid to speak up15:05
creihtk15:05
dhellmannare you going to be at the summit?15:06
creihtwell it will be kinda interesting since swift doesn't really use anything in oslo15:06
creihtyes I will be there15:06
dhellmanncreiht: well, maybe this is an opportunity to add some, then :-)15:07
creihthehe15:08
*** mriedem1 has quit IRC15:08
dimsdhellmann, nice sales pitch :)15:08
creihtwell from the swift perspective, we've always pushed for libraries15:08
dhellmanndims: hush, you're giving away the plan!15:08
creihtwe aren't very fond of the code copying stuff and the problems that can cause15:08
dhellmannindeed, we've definitely started to see scaling issues with the incubator approach15:09
creihtand after that it is mostly a matter of, if we use this library what benefits does it give us15:09
dhellmannand we have a fair amount of code that is stable enough to graduate, so things are progressing as planned15:09
creihtone of the design tennants of swift has always been to minimize dependencies15:10
* dhellmann nods15:10
creihtand I think that has really helped us tremendously15:10
*** ihrachyshka has quit IRC15:11
creihtfor example, we don't use any framework for the wsgi/routing/api stuff15:12
creihtturns out the swift api is pretty simple15:12
dimscreiht, does swift participate in the global requirements sync-ing? (pardon my ignorance)15:12
creihtwe played with a couple, and it really didn't make things that much better15:12
creihtdims: kinda15:12
creihtdims: Stuff isn't auto-synced15:13
dimsah thought so15:13
creihtbecause swift supports a broader swath of OS versions15:13
creihtauto syncing would break that as min versions are bumped up15:13
creihtfor example we will support Ubuntu Lucid15:14
creihtthe current global reqs will not work on lucid15:14
*** viktors has quit IRC15:14
dimsi see15:14
creihtplus we only have 7 requirements, so it is easy to keep up to date15:14
creihtwe manually update it when it makes sense15:15
creihtanother advantage of keeping a minimal set of dependencies15:15
creiht:)15:15
dimsright, so you would have to check which ones the oslo libs will drag in15:17
creihtright15:17
creihtand also another reason why we might be hesitant to bring an oslo lib in15:17
creihtwe also aren't likely to use an oslo lib just for the sake of using an oslo lib15:17
dhellmannyeah, the oslo libs will definitely be syncing automatically from the global requirements, but if we can update those global requirements to accommodate swift then we should keep that in mind as an option15:17
creihtthere has to be a compelling reason to do so15:18
dimsi hear ya15:18
creihtmost of oslo is also written from a nova style framework perspective15:18
creihtso a lot of it doesn't even make since for swift15:18
dhellmannsure15:18
creihtBut if there is a library that adds value, we will consider it15:19
dimscreiht, anything on the page look interesting? https://wiki.openstack.org/wiki/Oslo (section #2)15:20
creihtwell we already use hacking and pbr15:23
creihtwhen we get to the pluggable parts of the pluggable backends of swift I imagine we will explore stevedore15:24
creihtother than that, nothing obvious sticks out15:25
*** gcb_ has joined #openstack-oslo15:30
creihtI know there are some that would like see us use more of the oslo libraries15:30
dhellmannwe would rather share than re-invent, but there's no point to adding more libs just for the sake of it15:31
dims+115:32
creihtdhellmann: I appreciate that15:33
creihtI always get those questions each summit though15:34
creiht"Why doesn't swift use more of oslo"15:34
creihtthere's also a couple of areas of long standing differing of opinions on how some things should be done between swift and nova15:34
creihtthose aren't likely to change either15:35
dhellmannmy goal with these new libs is to make them focused enough that it's relatively easy to pick some of them and leave others out15:35
creihtbut the nova opinions made it into oslo15:35
dhellmannthere are interdependencies, but we want to minimize those15:35
dimscreiht, how about the opposite, what stuff would be valuable to others from swift?15:35
dhellmanngood question, I'm sure we'll find some parts of swift that would be good candidates for reuse elsewhere15:35
creihtwell some things that have come up before are our eventlet friendly memcache client15:36
creihtmaybe swob15:36
dhellmannthose both sound like they are worth looking at15:37
creihtother than that, we have some other eventlet helper things15:38
creihtif everyone else is fine with still using webob, then swob may not be that handy15:40
creihtwe just had so many bad experiences, we decided to write it15:40
creihtand it really isn't that much code15:40
dhellmannyeah, I think it's worth evaluating -- just going through the exercise can help us come up with guidelines for the sorts of things we would want to look at in other projects15:41
creihtcool15:41
dhellmanneven if we eventually say it should stay in swift, we'll have reasons for the decision15:41
creihtdhellmann: do the other projects still use paste.deploy to setup the wsgi pipelines? or is all of that in oslo.config?15:41
dhellmannI think they're still using paste, although more of that will go away as projects move to pecan15:42
creihthow does pecan setup the pipeline?15:42
creihtThat is one area of pain for us, and we want to look at other ways to handle it15:42
creihtmostly because swift is comprised of so much middleware now, that the pipeline gets a little crazy15:43
dhellmannit exposes a wsgi app, and the routing is done with the pecan controllers -- probably not a great solution for swift15:43
creihtyeah15:43
dhellmannyeah, the idea with the other projects is to just cut as much of the pipeline out as possible15:43
creihtwe just need something that creates the app from all of the components15:43
dhellmannceilometer uses a config option to disable the keystone middleware, for example15:43
creihtahh15:43
dhellmannand leaves the decision about whether to apply anything outside of the keystone middleware up to the deployer15:44
*** shakamunyi has joined #openstack-oslo15:44
dhellmannso you *can* use paste, if you need to, but it's not required15:44
*** shakamunyi has quit IRC15:44
creihtahh15:44
dhellmannyou can also deploy the ceilometer wsgi app directly15:44
creihtwe could use something a little smarter15:44
dhellmannyeah, there's probably a middle ground there15:45
creihtas a typical swift pipeline can have 20 items in it15:45
creihtand it starts getting confusing for users which ones have to go in front of others15:45
dhellmannis there any reason for the deployer to worry about that at all? why make it configurable?15:45
dhellmannhonest question, I'm not that familiar with swift internals15:46
creihtyeah valid question15:46
creihtpartly because some middleware is optional15:46
creihtsome middleware may be homegrown and they don't know where to put it15:46
dhellmannI think that was feedback from the folsom or grizzly summit, when we first started talking about dropping paste15:46
dhellmannah, home-grown stuff can be a challenge, yeah15:46
creihtdhellmann: https://github.com/openstack/swift/blob/master/etc/proxy-server.conf-sample#L7315:48
dhellmanngithub is slow today15:48
*** zhiyan is now known as zhiyan_15:50
dhellmannthat's quite a list15:50
creihtthat's what a standard pipeline looks like currrently15:50
creihtand my connection at this coffee shop is very laggy this morning15:50
creihtI imagine that we will move to something that is basically a helper that creates the pipeline, with some hooks to add your own pre and post15:51
creihtbut just wanted to bring it up in case anyone was working on something similar or new of something similar15:51
dhellmannI'm not aware of anything, but I wouldn't be surprised to learn that something like that already exists on pypi somewhere15:52
creihtyeah I haven't had a chance to poke yet15:53
creihtbut it is on my list15:53
creihtdhellmann: some of our middleware might be interesting15:53
creihtfor example rate limiting15:53
dhellmannyeah, definitely15:54
creihtright now it is a bit swift specific15:54
creihtbut works *amazingly* well15:54
creihtif others need that style of rate limiting, it could probably be more generalized15:54
creihtthe recon stuff is pretty cool, again swift specific, but general idea could be useful with other projects15:56
creihtif the profiling middleware gets in, that could be handy as an oslo project15:57
*** nacim has quit IRC15:58
dhellmannis that related to the rally stuff?15:58
dimsdhellmann, meeting now?16:01
dhellmanndims: we moved it to next week because today is a holiday for some folks16:01
* dhellmann checks that he sent that email16:02
creihtoh hehe... I was just wondering about the meeting :)16:02
dhellmanndims: http://lists.openstack.org/pipermail/openstack-dev/2014-April/032951.html16:02
creihtdhellmann: rally?16:02
dimsah thanks16:02
dhellmanncreiht: rally is a profiling tool boris-42 and some other folks at mirantis have been working on16:03
jd__ah I missed that too16:03
creihtahh don't think so16:03
jd__I was ready and hot for dat meeting16:03
dhellmannjd__: you could hang out and talk to us here, we miss you16:03
boris-42dhellmann actually not rally16:04
dhellmannnote that the time for the meeting next week is different, too16:04
creihtdhellmann: https://review.openstack.org/#/c/53270/16:04
dhellmannboris-42: am I mixing up projects again?16:04
boris-42dhellmann https://github.com/pboris/osprofiler16:04
dhellmannaha16:04
boris-42dhellmann btw seems like we have our devstack and 1 click demo16:04
dhellmannboris-42: excellent!16:05
boris-42dhellmann I would like to just finish work inside Rally16:05
boris-42dhellmann so we will be able to combine everything together16:05
boris-42dhellmann and show benchmarking + profiling togther16:05
dhellmannthat makes sense16:05
creihtdhellmann: I have a suggestion for: https://wiki.openstack.org/wiki/Oslo#Libraries16:08
creihtfor each library, it would be nice to have a link to the docs16:08
creihtfor example I've been clicking on all fo the links for oslo.config, and can't find any docs for it16:09
dhellmannwe do have some of them on http://docs.openstack.org/developer/openstack-projects.html, but adding them to the wiki makes sense, too16:09
creihtpypi page doesn't link to docs16:09
creihtthe launchpad page doesn't link to docs16:09
dhellmannyeah, we should add a link to the readme, too16:09
boris-42dhellmann btw are you going to approve design session about profiling stuff?16:10
boris-42dhellmann there is a lot of stuff to discuss=)16:12
boris-42dhellmann cause I think it's possible to make profiler in a such a way to be able to run it even on production clouds16:13
dhellmannboris-42: I need to review our available slots, but I do expect to include that one -- I know there's a lot of interest16:14
boris-42dhellmann yep cause without profiler you are not able to find the reason why we are getting so bad results..16:14
boris-42dhellmann sometimes it is clear by code, and sometimes not16:14
*** shakamunyi has joined #openstack-oslo16:18
*** bknudson has quit IRC16:32
openstackgerritMonty Taylor proposed a change to openstack-dev/pbr: Revert "Let git do mailmap mapping for us"  https://review.openstack.org/8862516:33
openstackgerritMonty Taylor proposed a change to openstack-dev/pbr: Revert "Let git do mailmap mapping for us"  https://review.openstack.org/8862716:40
*** gcb_ has quit IRC16:44
*** jogo is now known as flashgordon16:47
*** harlowja_away is now known as harlowja16:58
*** markmcclain has quit IRC16:59
openstackgerritMike Perez proposed a change to openstack-dev/hacking: Import Guideline Exception for gettextutils  https://review.openstack.org/8863217:04
openstackgerritMike Perez proposed a change to openstack-dev/hacking: Import Guideline Exception for gettextutils  https://review.openstack.org/8863217:05
*** arnaud__ has joined #openstack-oslo17:11
*** rpodolyaka1 has joined #openstack-oslo17:17
openstackgerritIsaku Yamahata proposed a change to openstack/oslo.messaging: _driver: implement unix domain support  https://review.openstack.org/7786217:21
openstackgerritIsaku Yamahata proposed a change to openstack/oslo.messaging: proxy: implement proxy server  https://review.openstack.org/7786317:21
*** bknudson has joined #openstack-oslo17:28
*** rpodolyaka1 has quit IRC18:39
*** arnaud__ has quit IRC18:46
*** rpodolyaka1 has joined #openstack-oslo18:57
*** dims is now known as dims-openstack18:58
*** ihrachyshka has joined #openstack-oslo19:07
*** arnaud__ has joined #openstack-oslo19:08
*** rpodolyaka1 has quit IRC19:18
*** dims-openstack is now known as dims19:24
*** shakayumi has joined #openstack-oslo19:26
*** shakamunyi has quit IRC19:27
*** tongli has joined #openstack-oslo19:35
*** rpodolyaka1 has joined #openstack-oslo19:44
openstackgerritJoshua Harlow proposed a change to openstack/taskflow: Start working on switch (conditional) - WIP  https://review.openstack.org/8741719:50
*** arnaud__ has quit IRC20:00
*** rpodolyaka1 has quit IRC20:39
*** arnaud__ has joined #openstack-oslo20:40
*** arnaud__ has quit IRC20:43
*** tongli has quit IRC20:49
*** harlowja is now known as harlowja_away20:50
*** jecarey has quit IRC20:55
*** arnaud__ has joined #openstack-oslo20:58
openstackgerritArnaud Legendre proposed a change to openstack/oslo-incubator: Add a Retry decorator in excutils  https://review.openstack.org/8869321:09
*** jecarey has joined #openstack-oslo21:19
*** harlowja_away is now known as harlowja21:41
*** ihrachyshka has quit IRC22:02
*** dims has quit IRC23:15
*** dims has joined #openstack-oslo23:42

Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!