| *** rlandy has quit IRC | 00:10 | |
| mordred | corvus: to be fair, clarkb does have 2 chilren unrelated to that change he could look at | 00:56 |
|---|---|---|
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul-jobs master: configure-pypi-mirrors: add initial role https://review.openstack.org/581165 | 01:10 |
| *** yolanda__ has joined #zuul | 01:33 | |
| *** yolanda_ has quit IRC | 01:36 | |
| *** jimi_|ansible has joined #zuul | 02:15 | |
| *** corvus_ has joined #zuul | 02:17 | |
| *** pleia2_ has joined #zuul | 02:17 | |
| *** corvus has quit IRC | 02:19 | |
| *** rcarrillocruz has quit IRC | 02:19 | |
| *** pleia2 has quit IRC | 02:19 | |
| *** Diabelko has quit IRC | 02:19 | |
| *** jimi|ansible has quit IRC | 02:19 | |
| *** eikke has quit IRC | 02:19 | |
| *** tobasco has quit IRC | 02:19 | |
| *** eikke has joined #zuul | 02:19 | |
| *** Diabelko has joined #zuul | 02:20 | |
| *** pleia2_ is now known as pleia2 | 02:20 | |
| *** eikke has quit IRC | 02:21 | |
| *** eikke has joined #zuul | 02:21 | |
| *** tobasco has joined #zuul | 02:22 | |
| *** rcarrillocruz has joined #zuul | 02:25 | |
| *** yolanda_ has joined #zuul | 02:28 | |
| *** yolanda__ has quit IRC | 02:31 | |
| *** corvus_ is now known as corvus | 03:36 | |
| *** corvus is now known as Guest78457 | 03:37 | |
| *** Guest78457 is now known as jeblair | 03:38 | |
| *** jeblair is now known as corvus | 03:43 | |
| *** ianychoi_ has quit IRC | 03:54 | |
| *** ianychoi has joined #zuul | 04:08 | |
| *** jappleii__ has quit IRC | 04:25 | |
| *** threestrands has joined #zuul | 04:25 | |
| *** yolanda__ has joined #zuul | 04:46 | |
| *** yolanda_ has quit IRC | 04:49 | |
| * SpamapS finally upgrading to zuul 3.1 .. I hope it doesn't break anything :-P | 05:13 | |
| tobiash | SpamapS: good luck ;) | 05:15 |
| SpamapS | Ya.. I kinda let my install get stale | 05:16 |
| SpamapS | I should probably re-sync zuul-jobs too | 05:17 |
| SpamapS | I'm a little worried the javascript has gotten weird again | 05:21 |
| SpamapS | would really love to not have a bunch of basically custom zuul deployment ansible :-P | 05:21 |
| tobiash | SpamapS: I think 3.1.0 is before the angular changes | 05:22 |
| tobiash | SpamapS: if you switch to current master and have a multi tenant setup you might need to tweak the reverse proxy settings a but | 05:23 |
| tobiash | bit | 05:23 |
| SpamapS | I said 3.1.. I meant master. ;) | 05:23 |
| SpamapS | We'll see. It's installing onto my stage env now. | 05:23 |
| tobiash | SpamapS: so if you have a white labeled setup you're fine, if you have a multi tenant setup you'll probably need to add a reverse proxy rule | 05:24 |
| SpamapS | Another thing I'd really love to see would be fast-forward job skips. If check has run for a job with the same git components.. just mark it SUCCESS pls. | 05:24 |
| SpamapS | tobiash: I am not white labeled, but have only one tenant.. for no reason other than I was too lazy to change all my links from /GoDaddy/foo to /foo. | 05:25 |
| tobiash | SpamapS: then you'll need something like this: http://paste.openstack.org/show/725383/ | 05:25 |
| SpamapS | Hm I might alreayd have something like that. | 05:26 |
| tobiash | (if you want a working tenant list) | 05:26 |
| SpamapS | ah, I may not care about that. :) | 05:27 |
| tobiash | SpamapS: but that only works with https://review.openstack.org/579418 | 05:27 |
| SpamapS | but it's good to know | 05:27 |
| tobiash | which is not yet merged | 05:27 |
| tobiash | that stuff is one of the reasons to hold back the next release containing the angular stuff | 05:28 |
| SpamapS | ah | 05:33 |
| SpamapS | well I'll see | 05:33 |
| SpamapS | can rewind to 3.1 if it gets too ugly ;) | 05:34 |
| SpamapS | we only have 10 users. ;) | 05:34 |
| openstackgerrit | Ian Wienand proposed openstack-infra/zuul-jobs master: [DNM] download links https://review.openstack.org/581204 | 06:05 |
| tobiash | we now have 1000 registered users (don't know how many are active atm) | 06:08 |
| *** threestrands has quit IRC | 06:30 | |
| *** hashar has joined #zuul | 06:34 | |
| *** yolanda_ has joined #zuul | 06:36 | |
| *** yolanda__ has quit IRC | 06:39 | |
| *** gtema has joined #zuul | 07:02 | |
| *** fbo|off is now known as fbo | 07:28 | |
| *** yolanda__ has joined #zuul | 07:59 | |
| *** yolanda__ is now known as yolanda | 07:59 | |
| *** yolanda_ has quit IRC | 08:02 | |
| *** electrofelix has joined #zuul | 08:03 | |
| *** hwoarang has joined #zuul | 08:13 | |
| *** hwoarang has quit IRC | 08:18 | |
| *** goern_ is now known as goern | 08:18 | |
| *** goern has quit IRC | 08:19 | |
| *** goern has joined #zuul | 08:19 | |
| *** sshnaidm|rover has quit IRC | 08:26 | |
| *** hwoarang has joined #zuul | 08:28 | |
| *** hwoarang has quit IRC | 08:28 | |
| *** hwoarang has joined #zuul | 08:28 | |
| *** sshnaidm|rover has joined #zuul | 08:32 | |
| openstackgerrit | Joshua Hesketh proposed openstack-infra/zuul master: Fix bindep for SUSE https://review.openstack.org/581252 | 08:32 |
| openstackgerrit | Joshua Hesketh proposed openstack-infra/zuul master: Fix github setup in zuul from scratch https://review.openstack.org/581253 | 08:32 |
| openstackgerrit | Joshua Hesketh proposed openstack-infra/zuul master: fix zuul from scratch user and group creation https://review.openstack.org/581254 | 08:32 |
| openstackgerrit | Joshua Hesketh proposed openstack-infra/zuul master: Add instructions for deploying zuul with openSUSE https://review.openstack.org/581255 | 08:32 |
| openstackgerrit | Joshua Hesketh proposed openstack-infra/zuul master: Include instructions on building static web components https://review.openstack.org/581256 | 08:32 |
| *** gtema has quit IRC | 08:37 | |
| *** toabctl has quit IRC | 08:48 | |
| *** toabctl has joined #zuul | 08:54 | |
| *** sshnaidm|rover has quit IRC | 09:04 | |
| *** sshnaidm|rover has joined #zuul | 09:05 | |
| sshnaidm|rover | how do I use zuul secret with files? I have a ssh private key that need to be used in jobs, what is good practice to inject it in jobs? | 09:22 |
| *** hwoarang has quit IRC | 09:23 | |
| tristanC | sshnaidm|rover: perhaps use the add-sshkey zuul-jobs role? | 09:24 |
| fbo | tristanC: left comment on 12929 | 09:31 |
| *** gtema has joined #zuul | 10:18 | |
| *** gtema has quit IRC | 10:35 | |
| *** gtema has joined #zuul | 10:46 | |
| *** hwoarang has joined #zuul | 10:47 | |
| odyssey4me | corvus and anyone else who may be able to act as a guide, I'm working on implementing the ability for nodepool to ignore the cloud quota if a config entry is set for a pool, as discussed here: http://eavesdrop.openstack.org/irclogs/%23zuul/%23zuul.2018-05-24.log.html#t2018-05-24T16:17:58 | 11:07 |
| odyssey4me | I'm trying to figure out exactly where to enact the use of the config entry and have tried it in the openstack handler's hasRemainingQuota and hasProviderQuota but that doesn't seem to be working... so I went down to the utils.py/non_negative and forced it to return True as a test, and it's still giving me quota issues... I must be barking up the wrong tree? | 11:09 |
| *** hwoarang has quit IRC | 11:19 | |
| *** hwoarang has joined #zuul | 11:20 | |
| *** hwoarang has quit IRC | 11:20 | |
| *** hwoarang has joined #zuul | 11:20 | |
| *** hwoarang has quit IRC | 11:21 | |
| *** hwoarang has joined #zuul | 11:22 | |
| Shrews | odyssey4me: i would think that adding a new config entry in nodepool/drivers/openstack/config.py for the provider, then checking this value in both of those mentioned methods would work | 11:24 |
| odyssey4me | Shrews: yep, doing that - but when it didn't take effect I starting trying to figure out why. And for some reason it seems that nothing at all is taking. | 11:24 |
| Shrews | odyssey4me: i think modifying the quota non_negative stuff itself is definitely the wrong direction | 11:25 |
| Shrews | odyssey4me: want to share your code and test with me? | 11:25 |
| odyssey4me | If I set both those methods to return True without checking anything, it still doesn't action the build and returns the error 'Declining node request 100-0000000908 because it would exceed quota' | 11:25 |
| odyssey4me | yeah, I can push up a review for now - it's incomplete but I'm happy to share where it's at | 11:26 |
| odyssey4me | lemme de-hack it quickly and push it up | 11:26 |
| Shrews | k | 11:26 |
| Shrews | odyssey4me: if you could add a test case for it testing the desired behavior (even though it may fail right now), that would be helpful | 11:28 |
| Shrews | but whatever you have now is fine, too | 11:29 |
| odyssey4me | ok, I can try that - I'm a bit of a noob, so it may be somewhat naiive | 11:29 |
| Shrews | ok, just push up what you have now | 11:29 |
| odyssey4me | lemme push up the modified code I was testing, then work out a test | 11:29 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack-infra/nodepool master: [WIP] Add ability to ignore provider quota for a pool https://review.openstack.org/581329 | 11:37 |
| odyssey4me | Shrews: ^ thanks | 11:38 |
| openstackgerrit | Joshua Hesketh proposed openstack-infra/zuul master: Add instructions for deploying zuul with openSUSE https://review.openstack.org/581255 | 11:42 |
| openstackgerrit | Joshua Hesketh proposed openstack-infra/zuul master: Include instructions on building static web components https://review.openstack.org/581256 | 11:42 |
| openstackgerrit | Tobias Henkel proposed openstack-infra/zuul master: Fix zuul startup with inexisting project template and gate https://review.openstack.org/579859 | 11:44 |
| openstackgerrit | Tobias Henkel proposed openstack-infra/zuul master: Tolerate missing project https://review.openstack.org/579872 | 11:44 |
| tobiash | corvus: stacked these ^ on top of your change 579997 because they conflict | 11:45 |
| Shrews | odyssey4me: so, couple of initial thoughts... ignoring quota should be an attribute of the provider (not pool), so in ProviderConfig | 11:50 |
| Shrews | odyssey4me: second thing, the has*Quota() methods can just check this value as the first check and just return True rather than doing *any* quota calculations | 11:51 |
| odyssey4me | on that second thought, I actually tried that - effectively just adding 'return True' in the method, but it still seems to be doing the same as always... that's what got me confused | 11:52 |
| odyssey4me | on the first thought - happy to do that, although the max-* settings are per pool, so I thought it was better suited to be there | 11:52 |
| Shrews | odyssey4me: that's where i'd need to see your test :) | 11:53 |
| odyssey4me | fair enough | 11:53 |
| odyssey4me | ok, I'll spend some time figuring out the test bit - at a glance it doesn't seem too hard, but as always the devil's in the details | 11:54 |
| Shrews | odyssey4me: hrm, would it make sense to have 2 different pools within the same provider and one ignores quota and the other doesn't? | 11:55 |
| Shrews | honest question... trying to think of the use case | 11:55 |
| Shrews | i suppose they could specify different AZs with different quota requirements... maybe pool is the right place here... | 11:56 |
| odyssey4me | good question, to which I don't really have an answer - my placement was more around proximity to the other bits which relate, rather than a use-case | 11:57 |
| Shrews | odyssey4me: yeah. just leave it in pool for now while i think about it a bit. i'm still pre-coffee and bfast | 11:57 |
| odyssey4me | we have found cases where boot-timeout/launch-timeout would be more useful in a pool than provider | 11:57 |
| odyssey4me | it's almost as if it'd be useful to have some settings that apply to both, where a pool's setting culd override the provider's setting | 11:58 |
| odyssey4me | *cloud | 11:58 |
| odyssey4me | bah *could | 11:58 |
| * odyssey4me gets coffee too :p | 11:58 | |
| *** elyezer has quit IRC | 12:02 | |
| *** elyezer has joined #zuul | 12:08 | |
| *** yolanda has quit IRC | 12:21 | |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack-infra/nodepool master: Update README to use python3 installation https://review.openstack.org/581348 | 12:31 |
| *** rlandy has joined #zuul | 12:35 | |
| *** yolanda has joined #zuul | 12:42 | |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack-infra/nodepool master: Update README to use python3 installation https://review.openstack.org/581348 | 12:42 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack-infra/nodepool master: [WIP] Add ability to ignore provider quota for a pool https://review.openstack.org/581329 | 12:52 |
| openstackgerrit | David Shrewsbury proposed openstack-infra/nodepool master: [WIP] Add ability to ignore provider quota for a pool https://review.openstack.org/581329 | 12:54 |
| Shrews | odyssey4me: oops, but the PS i pushed up adds a test | 12:55 |
| odyssey4me | Shrews: oh! thanks - didn't realise you were working on that | 12:55 |
| Shrews | odyssey4me: hopefully i didn't undo your changes | 12:55 |
| odyssey4me | happy to have that :) | 12:55 |
| Shrews | odyssey4me: it shows your changes worked as-is | 12:55 |
| odyssey4me | interesting, my test env must be funky then I guess | 12:58 |
| *** yolanda_ has joined #zuul | 13:00 | |
| *** yolanda has quit IRC | 13:02 | |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack-infra/nodepool master: [WIP] Add ability to ignore provider quota for a pool https://review.openstack.org/581329 | 13:03 |
| odyssey4me | Shrews: ^ adjusted based o your previous feedback | 13:04 |
| *** yolanda_ has quit IRC | 13:05 | |
| SpamapS | mmmmmmmmm new stuff Zuul version: 3.1.1.dev88 d7b835d | 13:10 |
| SpamapS | http://paste.openstack.org/show/725411/ | 13:14 |
| SpamapS | durn | 13:14 |
| tobiash | SpamapS: I think I saw a change from tristan to fix tha | 13:17 |
| tobiash | let me check | 13:18 |
| *** yolanda_ has joined #zuul | 13:18 | |
| SpamapS | I should always learn to search the review queue before opening up vim. | 13:18 |
| tobiash | SpamapS: https://review.openstack.org/577357 | 13:20 |
| tobiash | that might be your issue | 13:20 |
| SpamapS | Indeed, though the branch in question does actually have branch protections enabled. | 13:21 |
| tobiash | SpamapS: you could still open your vim and add a test case to that :-P | 13:21 |
| tobiash | SpamapS: does your zuul have permission to get that branch protection settings? | 13:21 |
| SpamapS | tobiash: it should, it has write access to the repo. | 13:22 |
| tobiash | SpamapS: on newer ghe without app auth the zuul techuser needs to be admin to get branch protection otherwise it will get the same result as it would be without branch protection | 13:22 |
| tobiash | SpamapS: ghe 2.12 changed that afaik | 13:23 |
| SpamapS | admin, not just write? | 13:23 |
| SpamapS | ahh I see | 13:23 |
| SpamapS | my robots are all just 'write' | 13:23 |
| tobiash | yes, since 2.12 you need admin to get branch protection | 13:23 |
| SpamapS | we just updated to 2.13 the other day :-P | 13:23 |
| tobiash | so that explains that | 13:24 |
| SpamapS | yep, upgrading our robots to admin on all the things | 13:24 |
| SpamapS | but yeah, still should return {} and not None. ;) | 13:25 |
| tobiash | SpamapS: I'm not opposed to this, my -1 on that fix is just for the missing test ;) | 13:25 |
| SpamapS | ok that seems to have gotten the jobs flowing again | 13:28 |
| SpamapS | hm, I have white borders on my terminal watcher | 13:29 |
| SpamapS | whats up with that? | 13:29 |
| tobiash | SpamapS: that's the new angular design | 13:30 |
| tobiash | the terminal watcher is its own component now | 13:30 |
| tobiash | but you should ask tristanC and mordred for the details ;) | 13:30 |
| SpamapS | mmk | 13:30 |
| mordred | SpamapS: javascript | 13:31 |
| SpamapS | wrote a song about it | 13:31 |
| SpamapS | like to hear it? here it goes | 13:31 |
| * tobiash listens | 13:31 | |
| SpamapS | mmmmmmmm hmmmm hmmmm mmmmmm.. we got ourselves a scripting language. It runs in all the browsers. mmmmhmmm mhmmmmmmmmmmm | 13:31 |
| SpamapS | thank you thank you very much | 13:31 |
| mordred | that's a great song | 13:31 |
| SpamapS | wrote it myself | 13:31 |
| mordred | you should totally go into the songwriting business | 13:32 |
| mordred | maybe justin bieber will record your songs | 13:32 |
| SpamapS | he already does | 13:32 |
| Shrews | SpamapS: ah, you are so living in colour | 13:33 |
| Shrews | s/colour/color/ ... i went all britishy there | 13:33 |
| mordred | Shrews: you're not british? | 13:34 |
| SpamapS | Shrews: SpamapS don't play dat. | 13:34 |
| SpamapS | ahh fun, new ansible version is breaking some of my jobs. How quaint. | 13:37 |
| *** adam_g has quit IRC | 13:38 | |
| *** adam_g has joined #zuul | 13:38 | |
| tobiash | odyssey4me, Shrews: regarding that ignore quota thing you might want to think about how that interacts with https://review.openstack.org/536930 | 13:39 |
| tobiash | odyssey4me, Shrews: I think that this should not be a problem but it's worth thinking about that | 13:40 |
| Shrews | tobiash: good point. i'm not sure how that plays in the rax environment odyssey4me is describing | 13:42 |
| odyssey4me | tobiash: yes, in principle that becomes a nice fallback in case the nodepool max-servers setting is actually too high (or something else is using quota)... I'm looking through the patch itself to see if that perhaps might explain why my patch isn't working :p | 13:42 |
| Shrews | odyssey4me: the way you have the code now, max-servers is not even considered (except for disabling a pool) | 13:43 |
| odyssey4me | Shrews: oh dear, ok, let me revisit that | 13:43 |
| Shrews | (sort of relates to my comment i just left on the commit message) | 13:43 |
| tobiash | odyssey4me: the intention is to put the handler into pause mode after catching a quota related launch exception, after that it waits until quota is already available (which would then be probably immediately) | 13:43 |
| tobiash | odyssey4me: so you might be hammering the cloud until it works if you set the wrong max-servers | 13:44 |
| odyssey4me | hmm, yeah - I see that the patch intentionally allows the request to be re-fulfilled in the same provider... why not allow it to be fulfilled by *any* provider? | 13:47 |
| tobiash | odyssey4me: because it can be part of a multi node set which needs to be fulfilled by the same provider | 13:47 |
| odyssey4me | aha, I see | 13:47 |
| odyssey4me | will that then keep retrying until the full node set is available? or will the full node set be aborted and is open to fulfillment elsewhere? | 13:48 |
| *** jimi_|ansible is now known as jimi|ansible | 13:48 | |
| tobiash | it will keep retrying | 13:48 |
| tobiash | it just aborts the failing node | 13:49 |
| odyssey4me | ok, good to know | 13:49 |
| SpamapS | https://docs.ansible.com/ansible/2.5/porting_guides/porting_guide_2.5.html#fixed-handling-of-keywords-and-inline-variables ... durn... some of my ansible used {{ name }} :-P | 13:58 |
| * SpamapS fixed it I think | 13:59 | |
| *** yolanda_ has quit IRC | 14:03 | |
| jhesketh | tristanC: ping | 14:08 |
| SpamapS | The multi-node set requiring to be handled by the same provider and same AZ is something I've wanted to revisit for a while. | 14:09 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack-infra/nodepool master: [WIP] Add ability to ignore provider quota for a pool https://review.openstack.org/581329 | 14:09 |
| SpamapS | I think in some cases a job should be able to express that it doesn't care about that. | 14:09 |
| tobiash | SpamapS: at least as a nodepool operator I also don't care about AZ in our cloud | 14:16 |
| SpamapS | yeah, at some point I had a patch prepared that ignored AZ-stickiness. | 14:17 |
| SpamapS | don't remember where it went | 14:17 |
| tobiash | SpamapS: did that land? | 14:17 |
| SpamapS | no | 14:17 |
| SpamapS | I dunno if I even git-review'ed it | 14:17 |
| SpamapS | https://review.openstack.org/#/c/546175/ | 14:18 |
| *** yolanda_ has joined #zuul | 14:18 | |
| tobiash | oh I even reviewed this :) | 14:19 |
| SpamapS | yah.. thread found.. | 14:20 |
| SpamapS | but I actually found that the main reason I had laggy job starts was my zk node being too busy | 14:20 |
| SpamapS | once I moved it to a dedicated VM even the lag to start a few more nodes wasn't as noticable, so this task got pushed to the bottom of the stack | 14:21 |
| tobiash | but I still had issues this week wher my cloud added new nodes with a different AZ and then changed that and I got node failures because nodepool tried to spawn nodes in an inexisting AZ | 14:21 |
| SpamapS | ah that's a bummer. ;) | 14:22 |
| tobiash | I guess we need to handle this error and in my case I wouldn't even care about AZs | 14:22 |
| SpamapS | we don't add AZ's ever.. we just expand the existing ones. | 14:22 |
| tobiash | yeah that probably was a deployment issue that was fixed after the deployment and after nodepool noticed that there is an additional AZ | 14:23 |
| SpamapS | Our cloud is also kinda weird with AZ's | 14:24 |
| SpamapS | you *have* to pick one | 14:24 |
| SpamapS | We built a routed networks openstack, before neutron had routed networks support. | 14:24 |
| SpamapS | So az is how you select your over-arching network. | 14:24 |
| tobiash | oh that sounds like fun | 14:25 |
| SpamapS | yyyyyyeaaahhh | 14:25 |
| SpamapS | Our newer control planes just use routed networks. | 14:25 |
| SpamapS | we also do this weird thing where we bind floating ips directly on the hosts and then have an l3 driver that tells the switch how to route the floating IP to the fixed. It's nice because there's no soft routers.. but it also means we can't use stock images because our images have to know how to bind a floating IP. :-P | 14:27 |
| odyssey4me | Shrews: weird, adjusted https://review.openstack.org/#/c/581329/5/nodepool/driver/openstack/handler.py to now just ignore the cloud quota, but my live test still seems to be hitting quota and completely ignoring the code | 14:30 |
| *** acozine1 has joined #zuul | 14:31 | |
| odyssey4me | I'm still hitting https://github.com/openstack-infra/nodepool/blob/master/nodepool/driver/__init__.py#L527 somehow. | 14:32 |
| *** EmilienM is now known as EvilienM | 14:42 | |
| *** acozine1 has quit IRC | 14:48 | |
| *** acozine1 has joined #zuul | 14:48 | |
| mordred | SpamapS: why are you using floating ips for that use case? | 14:55 |
| mordred | SpamapS: like, you just described normal neutron ports | 14:55 |
| mordred | SpamapS: (honest question - not a troll) | 14:56 |
| tobiash | corvus: circling back to the semi full reconfiguration possibility. What method for triggering this would you prefer? There are a number of possibilities like timer, inotify, SIGUSRx, control socket. | 15:14 |
| tobiash | corvus: I'd vote for control socket | 15:14 |
| corvus | tobiash: yes, and i think that will be more compatible with fbo's use of a script instead of a file | 15:15 |
| tobiash | corvus: I just noticed that 579997 is semantically conflicting with the job contamination fix | 15:20 |
| tobiash | I added both to my staging branch and that broke every test | 15:21 |
| corvus | tobiash: you wrote about the fix in a comment though, right? it looks straightforward... | 15:22 |
| tobiash | corvus: yes, it's easy :) | 15:22 |
| corvus | tobiash: job contamination is more important, let's land it first | 15:22 |
| tobiash | k | 15:23 |
| odyssey4me | Shrews never mind, I found the bug - it was the squishy bit between the keyboard and chair (me) | 15:42 |
| Shrews | :) | 15:44 |
| mhu | pabelanger, SpamapS do you mind if I take a look at/possibly take over https://review.openstack.org/#/c/95035/ ? it's a nice feature to have, I'd like to at least attempt to rebase that change | 15:49 |
| openstackgerrit | Matthieu Huin proposed openstack-infra/zuul master: Add a dequeue command to zuul client https://review.openstack.org/95035 | 15:52 |
| pabelanger | mhu: yes, you'll need to port unit test to zuulv3 layout I believe | 15:57 |
| mhu | pabelanger, thx, and yeah I'll do that in my next patch | 15:58 |
| *** sambetts has joined #zuul | 16:01 | |
| sambetts | Hi zuul, having a problem with zuul-web, it looks like the static files should have been included in the pypi package when it was built, however they are missing | 16:02 |
| sambetts | are there any docs on how to manually build and add the files without reinstalling zuul from source | 16:08 |
| sambetts | ? | 16:08 |
| pabelanger | sambetts: https://zuul-ci.org/docs/zuul/developer/javascript.html should help | 16:10 |
| pabelanger | but I want to say JS missing from pypi might be a bug? | 16:10 |
| sambetts | yeah I think it seems to be, the MANIFEST in the repo looks like it is trying to include it, but downloading the tar.gz from pypi its missing | 16:11 |
| sambetts | and after installing it, its not there | 16:11 |
| sambetts | so zuul-web for the API works, but the dashboards is completely dead | 16:12 |
| *** sshnaidm|rover is now known as sshnaidm|bbl | 16:12 | |
| sambetts | s/is/are/ | 16:12 |
| pabelanger | I know in openstack there has been some history with distro packagers about installing config files with pip, but don't think an issue for static html file | 16:12 |
| *** hashar is now known as hasharDinner | 16:17 | |
| sambetts | yeah they aren't config files they are static assets, it looks like someone forgot to do the npm build before creating the sdist | 16:21 |
| SpamapS | mhu: go for it! | 16:32 |
| *** gtema has quit IRC | 16:35 | |
| *** hwoarang has quit IRC | 16:42 | |
| *** fbo is now known as fbo|off | 16:44 | |
| odyssey4me | Shrews: how does one go about running the unit tests? are there any special requirements (ie could I run it on a macbook)? is it just tox -e foo, or more? | 16:48 |
| tobiash | odyssey4me: on a macbook is more difficult | 16:49 |
| Shrews | odyssey4me: nodepool follows (pretty much) the same PTI as OpenStack projects | 16:49 |
| tobiash | odyssey4me: with some tweaks you can get away using docker but that's awfully slow on mac | 16:50 |
| Shrews | odyssey4me: tox -epy35 (or -epy36 if you have that) | 16:50 |
| Shrews | odyssey4me: you need to start a zookeeper process | 16:50 |
| Shrews | but that's all you need | 16:50 |
| tobiash | odyssey4me: so I personally have a tox wrapper that rsyncs the source to a linux vm and runs tox there | 16:50 |
| Shrews | yeah, we don't really try to support a mac environment | 16:51 |
| Shrews | i tried once long ago... it became... tedious | 16:51 |
| odyssey4me | ok, let me update that README again with a bit more detail | 16:51 |
| tobiash | it's not just not trying, it's just not possible right now (natively) | 16:52 |
| *** sambetts is now known as sambetts|afk | 16:53 | |
| tobiash | some time ago I think gear was not compatible because it relies on some linux only socket flags | 16:53 |
| Shrews | tobiash: i don't doubt it | 16:53 |
| Shrews | epoll vs kqueue, etc | 16:53 |
| tobiash | Shrews: ah yes, epoll was it afaik | 16:54 |
| Shrews | that's when i gave up | 16:54 |
| tobiash | oh wow, at least pep8 on nodepool is working on mac | 16:54 |
| odyssey4me | Why does the README currently include cloning project-config? | 16:55 |
| *** acozine1 has quit IRC | 16:55 | |
| Shrews | odyssey4me: that whole doc could probably stand a revision now. even the 1st paragraph needs updated | 16:55 |
| odyssey4me | am I missing something? I don't see a py35/py36 env defined in tox.ini | 16:57 |
| Shrews | honestly, i'd rather strip most of the contents out of README and put them in the docs proper. The testing stuff could go in the developer docs | 16:57 |
| Shrews | odyssey4me: it's built in to tox | 16:57 |
| odyssey4me | ah ok | 16:59 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack-infra/nodepool master: Update README to use python3 installation https://review.openstack.org/581348 | 16:59 |
| odyssey4me | ok, I'll revise it all tomorrow and move it into the dev docs instead. | 17:00 |
| odyssey4me | thanks for your help today folks - https://review.openstack.org/581329 appears to be working as intended... so now it's down to polishing, better validations and tests | 17:01 |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul-jobs master: Switch to centos-release-openstack-queens for ovs https://review.openstack.org/580518 | 17:02 |
| odyssey4me | Shrews: FYI you were kinda right - it's a bit pointless to ignore the quota for a single pool, well for this use-case certainly... If you think it better to pull it back into the provider, please let me know in review and I'll adjust it tomorrow | 17:03 |
| pabelanger | ianw: ^updated to address comments | 17:03 |
| openstackgerrit | Logan V proposed openstack-infra/zuul master: encrypt_secret: Allow file scheme for public key https://review.openstack.org/581429 | 17:17 |
| SpamapS | mordred: the thing that makes it not like neutron ports is routing | 17:23 |
| SpamapS | mordred: this isn't a single L2. There are many L2's that the floating ips can move between. | 17:24 |
| mordred | SpamapS: ok. I believe you. :) | 17:26 |
| SpamapS | mordred: good, I don't want to trigger your floating ip rage by showing you how it gets used. ;) | 17:28 |
| *** electrofelix has quit IRC | 17:34 | |
| tobiash | Shrews, odyssey4me: just tried to run 'tox -e py36' in nodepool on mac and it worked :) | 17:43 |
| odyssey4me | tobiash: ORLY? did you have to install a zookeeper instance first? | 17:44 |
| tobiash | brew install zookeeper | 17:44 |
| tobiash | zkServer start | 17:45 |
| tobiash | and I needed to install gnu-getopt into the system | 17:45 |
| Shrews | odyssey4me: tobiash: i would not use brew | 17:45 |
| tobiash | Shrews: what else do you recommend? | 17:46 |
| Shrews | unless it doesn't start the service by default | 17:46 |
| tobiash | it doesn't | 17:46 |
| Shrews | oh, that's probably ok then. it's not something you want to keep running on a laptop | 17:46 |
| tobiash | it told me that I can activate the service or start it manually | 17:46 |
| Shrews | i just download it from apache.org and start it from the download | 17:47 |
| tobiash | sure that's possible too | 17:47 |
| tobiash | I was just lazy and didn't even expect that it could work | 17:48 |
| odyssey4me | looks like gnu-getopt is brew installable too | 17:48 |
| tobiash | but I'm sure zuul won't work because of gearman and bwrap | 17:48 |
| tobiash | yes | 17:49 |
| tobiash | nodepool needs it in the path during the tests | 17:49 |
| tobiash | so it looks like ripping out gearman during the v3 transition was the only thing needed to get the nodepool tests working on mac | 17:51 |
| SpamapS | I wouldn't really trust a test run on a mac. ;) | 18:23 |
| tobiash | SpamapS: I trust zuul :) | 18:25 |
| SpamapS | AFAIK zuul only really works on Linux. :) | 18:26 |
| tobiash | it just makes development on mac a little easier | 18:26 |
| tobiash | zuul probably won't never work on mac | 18:26 |
| tobiash | corvus: I think we should add both full reconfiguration and semi full reconfiguration to the command socket | 18:35 |
| tobiash | corvus: do you have whishes for the naming of the commands? | 18:35 |
| tobiash | corvus: the first commands that came to my mind were 'reconfigure' and 'reload' but that's probably too ambigious | 18:37 |
| tobiash | maybe 'reconfigure' and 'reconfigure-tenants' ? | 18:37 |
| corvus | tobiash: maybe "full-reconfigure" for the current clear-cache-and-reconfigure. then "reconfigure" for the new reload-tenant-config-but-use-cache option. we might want to reserve "tenant-reconfigure" or similar if we ever add a manual trigger for the current "tenant reconfiguration" which is automatically triggered by changes landing. | 18:53 |
| tobiash | corvus: thanks, that sounds good | 18:54 |
| *** rlandy is now known as rlandy|brb | 18:56 | |
| pabelanger | mordred: corvus: I'd like to see if I could spend some time on implementing: http://lists.openstack.org/pipermail/openstack-infra/2017-October/005606.html Change publication interface to be directories on node, not executor We just had a ssh key in rdoproject leak due to misunderstanding of how secrets work. And in this case, the job it self doesn't really need access to it if we properly setup publication | 19:17 |
| pabelanger | interfaces | 19:17 |
| mordred | pabelanger: awesome | 19:17 |
| corvus | ++ | 19:18 |
| mordred | pabelanger: I have the beginnings implemented - I just keep failing at pushing it through | 19:18 |
| corvus | pabelanger: but also, perhaps privately when you have a minute, i'd like to hear about the misunderstanding that caused that to see if there's any improvements we can make | 19:18 |
| mordred | pabelanger: https://review.openstack.org/#/q/topic:zuulv3-output | 19:19 |
| pabelanger | corvus: sure, https://review.rdoproject.org/r/14754/ was the change added this morning, i didn't have a chance to review until a few hours after it merged | 19:19 |
| pabelanger | corvus: for legacy reasons jobs still configured to rsync directly from a nodepool node to images.r.o, so key was missing when we ran zuul-migrate | 19:20 |
| pabelanger | however, because key was writen to disk, anybody could echo it | 19:21 |
| mordred | pabelanger: ah - fun | 19:21 |
| mordred | pabelanger: I think that's ACTUALLY the whole implementation above - other than stepping it through and fixing the bugs that are sure to be there since I wrote it | 19:21 |
| mordred | pabelanger: but I *believe* it's got all the major pieces | 19:22 |
| pabelanger | so, I am not sure how better to protect that from zuul POV. Initial thoughts were maybe set final: true by default if using a secret, and force operator to disable it or warn that allow-projects is missing... but that feels clunky | 19:22 |
| pabelanger | mordred: okay, let me review | 19:23 |
| mordred | pabelanger: I'm still down a couple of different rabbit holes - so if you need to fix anything or pushanything forward, please don't block on me - and feel free to take over any of the pieces | 19:23 |
| pabelanger | sure | 19:23 |
| mordred | pabelanger: I'm also happy to help - but it might be a few days before I'm out of my current hole | 19:23 |
| pabelanger | np, I think pushing on this for openstack first makes sense, then we can see how to get other zuul onboard | 19:24 |
| *** rlandy|brb is now known as rlandy | 19:27 | |
| mordred | pabelanger: ++ | 19:27 |
| *** hasharDinner is now known as hashar | 19:32 | |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul master: Add job to build container images using pbrx https://review.openstack.org/580160 | 19:42 |
| *** sshnaidm|bbl has quit IRC | 19:45 | |
| Shrews | mordred: that ^^ is pretty neat | 19:51 |
| *** sshnaidm|bbl has joined #zuul | 19:52 | |
| Shrews | so pbrx is just dox on steroids, eh? | 19:53 |
| mordred | Shrews: yeah - I guess it kinda is :) | 19:53 |
| mordred | Shrews: speaking of - wanna be core on pbrx? | 19:53 |
| * Shrews knew he should have kept his mouth shut | 19:54 | |
| mordred | Shrews: too slow - there you go | 19:54 |
| mordred | thanks for volunteering | 19:54 |
| Shrews | yay? | 19:54 |
| *** hwoarang has joined #zuul | 20:08 | |
| *** hwoarang has quit IRC | 20:10 | |
| *** hwoarang has joined #zuul | 20:11 | |
| mordred | Shrews, corvus: ZOMG - the job worked! http://logs.openstack.org/60/580160/12/check/zuul-build-container-images/a675a7f/job-output.txt.gz | 20:25 |
| corvus | mordred: i never doubted it! | 20:25 |
| mordred | corvus: golly I sure did | 20:25 |
| Shrews | wasn't the job just echo'ing the command though? | 20:26 |
| corvus | i'm delighted that there is a zuul-bwrap container | 20:26 |
| Shrews | https://review.openstack.org/#/c/580160/12/Dockerfile.zuul-executor line 9 | 20:27 |
| Shrews | oh nope, nm | 20:27 |
| Shrews | there was more to it :) | 20:27 |
| mordred | corvus: right?' | 20:27 |
| Shrews | yay | 20:27 |
| corvus | mordred: atm the images are just lost on the build node, right? they aren't uploaded anywhere (not even logs or tarball) right? | 20:28 |
| mordred | corvus: that is correct | 20:28 |
| mordred | corvus: next step - upload them somewhere | 20:28 |
| corvus | sounds promising | 20:29 |
| mordred | corvus: also, need to get zuul-dashboard done | 20:29 |
| mordred | since it is not really covered yet | 20:30 |
| corvus | SpamapS, jlk: have time/interest in reviewing a github change? https://review.openstack.org/580218 | 20:30 |
| corvus | tristanC: https://review.openstack.org/577357 could use your attention | 20:35 |
| *** acozine1 has joined #zuul | 20:36 | |
| mordred | corvus: fwiw, running 'pbrx build-images' in your zuul source tree should cause you to have docker images for each of those things locally if you wanted to play with them | 20:38 |
| Shrews | hopefully you can get pbrx to run! | 20:50 |
| * Shrews still puzzling through that one | 20:51 | |
| mordred | corvus: we may have just discovered that python3 will let you import things from directories without __init__.py files | 20:51 |
| mordred | but python2 will not | 20:52 |
| Shrews | mordred: i have NO idea how your test worked, but i'm just shrugging it off as "some-other-python-weirdness" that i don't want to get into | 20:53 |
| mordred | Shrews: me either | 20:55 |
| corvus | is anyone other than tobias using windows build nodes? | 20:59 |
| openstackgerrit | Merged openstack-infra/zuul master: Fix GitHub branch protection test https://review.openstack.org/580966 | 20:59 |
| openstackgerrit | Merged openstack-infra/zuul master: Fix secret example yaml https://review.openstack.org/580039 | 20:59 |
| jlk | corvus: looking | 21:01 |
| mordred | corvus: notit | 21:02 |
| *** acozine1 has quit IRC | 21:02 | |
| openstackgerrit | Merged openstack-infra/zuul master: Fix logging of clone url https://review.openstack.org/580053 | 21:04 |
| openstackgerrit | Matthieu Huin proposed openstack-infra/zuul master: Add a dequeue command to zuul client https://review.openstack.org/95035 | 21:11 |
| *** yolanda__ has joined #zuul | 21:18 | |
| corvus | tobiash: i'm leaving https://review.openstack.org/578230 for you to +A | 21:20 |
| *** yolanda_ has quit IRC | 21:21 | |
| *** hwoarang has quit IRC | 21:24 | |
| openstackgerrit | Merged openstack-infra/zuul master: Reduce requests to github when listing project branches https://review.openstack.org/580218 | 21:24 |
| *** hwoarang has joined #zuul | 21:25 | |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul-jobs master: Add role to ensure per-node output dirs exist https://review.openstack.org/511823 | 21:52 |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul-jobs master: Add role to fetch output from nodes https://review.openstack.org/511843 | 21:52 |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul-jobs master: Update upload-logs to process docs as well https://review.openstack.org/511853 | 21:52 |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul-jobs master: Upload fetched artifacts to log server https://review.openstack.org/511858 | 21:52 |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul-jobs master: Only publish artifacts when zuul.change is not defined https://review.openstack.org/511859 | 21:52 |
| openstackgerrit | Merged openstack-infra/zuul master: executor: change execution log to INFO https://review.openstack.org/578704 | 22:16 |
| openstackgerrit | Merged openstack-infra/zuul-jobs master: Switch to centos-release-openstack-queens for ovs https://review.openstack.org/580518 | 22:17 |
| *** yolanda_ has joined #zuul | 22:29 | |
| *** hashar has quit IRC | 22:31 | |
| *** yolanda__ has quit IRC | 22:32 | |
| openstackgerrit | Merged openstack-infra/zuul master: Add test facility to add file contents in github tests https://review.openstack.org/580948 | 22:32 |
| openstackgerrit | Merged openstack-infra/zuul master: Fix job contamination by unmerged change https://review.openstack.org/580949 | 22:32 |
| openstackgerrit | Merged openstack-infra/zuul master: Require tenant in Pipeline constructor https://review.openstack.org/581129 | 22:32 |
| *** rlandy is now known as rlandy|bbl | 22:44 | |
| tristanC | jhesketh: pong? | 23:36 |
| openstackgerrit | Logan V proposed openstack-infra/zuul master: Fix incorrect API URL for project key https://review.openstack.org/580034 | 23:41 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!