| *** bstinson has quit IRC | 00:54 | |
| *** bstinson has joined #zuul | 00:54 | |
| *** harlowja has quit IRC | 03:08 | |
| *** rlandy|bbl is now known as rlandy | 03:30 | |
| *** rlandy has quit IRC | 04:35 | |
| *** threestrands_ has joined #zuul | 05:09 | |
| *** threestrands_ has quit IRC | 05:09 | |
| *** threestrands_ has joined #zuul | 05:09 | |
| *** threestrands has quit IRC | 05:09 | |
| *** threestrands_ has quit IRC | 05:10 | |
| *** threestrands_ has joined #zuul | 05:11 | |
| *** threestrands_ has quit IRC | 05:11 | |
| *** threestrands_ has joined #zuul | 05:11 | |
| *** ankkumar has joined #zuul | 05:14 | |
| *** harlowja has joined #zuul | 05:15 | |
| *** threestrands has joined #zuul | 05:15 | |
| *** threestrands has quit IRC | 05:15 | |
| *** threestrands has joined #zuul | 05:15 | |
| *** threestrands_ has quit IRC | 05:18 | |
| *** threestrands has quit IRC | 05:28 | |
| *** ankkumar_ has joined #zuul | 05:52 | |
| *** ankkumar has quit IRC | 05:54 | |
| *** harlowja has quit IRC | 06:14 | |
| *** ankkumar__ has joined #zuul | 06:40 | |
| *** ankkumar_ has quit IRC | 06:44 | |
| openstackgerrit | Merged openstack-infra/zuul master: Invert executor ram statsd metric https://review.openstack.org/544742 | 06:57 |
|---|---|---|
| *** threestrands has joined #zuul | 07:09 | |
| *** threestrands has quit IRC | 07:09 | |
| *** threestrands has joined #zuul | 07:09 | |
| *** jpena|off is now known as jpena | 08:26 | |
| *** jaianshu has joined #zuul | 08:50 | |
| *** sshnaidm|off is now known as sshnaidm|ruck | 08:55 | |
| *** hashar has joined #zuul | 09:01 | |
| *** ankkumar__ has quit IRC | 09:14 | |
| *** threestrands_ has joined #zuul | 09:21 | |
| *** threestrands has quit IRC | 09:21 | |
| *** hashar is now known as hasharLunch | 10:45 | |
| *** tosky has joined #zuul | 10:48 | |
| *** jpena is now known as jpena|lunch | 11:52 | |
| *** elyezer has quit IRC | 12:25 | |
| *** elyezer has joined #zuul | 12:38 | |
| *** jaianshu has quit IRC | 13:02 | |
| *** markush1 has joined #zuul | 13:05 | |
| *** markush1 has left #zuul | 13:07 | |
| *** markush1 has joined #zuul | 13:22 | |
| *** hasharLunch is now known as hashar | 13:22 | |
| *** markush1 is now known as maho | 13:23 | |
| *** rlandy has joined #zuul | 13:32 | |
| *** jpena|lunch is now known as jpena | 13:37 | |
| openstackgerrit | Simon Westphahl proposed openstack-infra/zuul master: Allow using remote refs to find commits for change https://review.openstack.org/544964 | 13:48 |
| swest | tobiash: corvus: ^ proposal base on http://lists.zuul-ci.org/pipermail/zuul-discuss/2018-February/000023.html | 13:54 |
| *** electrofelix has joined #zuul | 14:19 | |
| *** threestrands_ has quit IRC | 14:39 | |
| *** dtruong has quit IRC | 14:40 | |
| *** dtruong has joined #zuul | 14:41 | |
| SpamapS | heh, running into an interesting problem right now | 14:52 |
| SpamapS | I want to move a job from an untrusted repo, to a trusted repo, so I can add upload creds to it | 14:53 |
| SpamapS | (so I can have many other repos upload built rpms) | 14:53 |
| SpamapS | but the only way I can do that is to remove the job from all the other repos first | 14:53 |
| SpamapS | because otherwise I'd need a cycle of Depends-On's to get zuul to parse the configs that both remove, and add it. | 14:54 |
| SpamapS | https://storyboard.openstack.org/#!/story/2001543 | 14:58 |
| SpamapS | hopefully I'm wrong | 14:58 |
| SpamapS | and somebody can just write how to do it in there, and we can turn it into a doc thing | 14:58 |
| dmsimard | SpamapS: sounds like a similar process to rename jobs | 15:23 |
| dmsimard | which requires ~3 patches | 15:23 |
| dmsimard | SpamapS: for example https://review.openstack.org/#/q/topic:base-rename | 15:23 |
| *** rbergeron has quit IRC | 16:10 | |
| *** rbergeron has joined #zuul | 16:10 | |
| *** rlandy is now known as rlandy|rover | 16:11 | |
| *** sshnaidm|ruck is now known as sshnaidm|notruck | 16:12 | |
| *** sshnaidm|notruck is now known as sshnaidm|mitsubi | 16:13 | |
| *** sshnaidm|mitsubi is now known as sshnaidm | 16:13 | |
| mordred | SpamapS: also, you can add upload creds to the job without moving it to a trusted repo | 16:36 |
| openstackgerrit | Merged openstack-infra/zuul master: Add zuul from scratch doc https://review.openstack.org/542518 | 16:39 |
| openstackgerrit | Merged openstack-infra/zuul master: Reclassify some debug log lines as info https://review.openstack.org/543044 | 16:39 |
| corvus | in exceptional situtations, you might be able to abuse the shadow feature to perform a move with fewer changes | 16:40 |
| *** tosky has quit IRC | 16:42 | |
| *** hashar is now known as hasharAway | 16:53 | |
| *** electrofelix has quit IRC | 16:55 | |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Add host/group vars https://review.openstack.org/544562 | 16:57 |
| SpamapS | mordred: I need the same upload creds for ~5 repos. | 17:03 |
| SpamapS | mordred: I suppose I could add the secret to each one with a different name, but it's really a shared infrastructure (a yum repo to upload to) | 17:03 |
| SpamapS | corvus: if I could use shadow and not remove all references to it, that would be great. | 17:04 |
| corvus | SpamapS: we haven't tried this yet, because so far, it's just been less hassle to do the 3- or 5-point-turn. but theoretically, you could configure your old repo to shadow the new repo, then add the job to the new repo, remove the job from the old one, then drop the shadow config. | 17:06 |
| corvus | shadow is a tenant config change, so requires a full reconfiguration to add/remove. it's still a hassle, but it's a *different* hassle that may be worth it if the alternative is to touch lots of repos. | 17:06 |
| SpamapS | corvus: that would, indeed, be a better way. :) | 17:07 |
| SpamapS | Because I can see us losing control of the repos that use a job. | 17:07 |
| corvus | SpamapS: let me know how it works :) | 17:08 |
| SpamapS | Unfortunately I already did the dance. So I'd have to simulate it and ENOTIME :-/ | 17:08 |
| SpamapS | BTW, the error messages were fantastic. | 17:09 |
| SpamapS | I was just blindly submitting patches and zuul kept telling me what was wrong nicely. | 17:10 |
| *** harlowja has joined #zuul | 17:19 | |
| mordred | SpamapS: just to circle back - you can put a secret on a job and have that shared job be used by 5 different repos without that job being in a trusted repo | 17:24 |
| mordred | SpamapS: which is mostly me wanting to poke at making sure we're not missing something in terms of communication about that | 17:25 |
| SpamapS | mordred: my experience is the opposite. If I do that, the secret is not exposed to the jobs in the other repos. | 17:32 |
| SpamapS | Entirely possible there's some subtle thing I missed while doing that. | 17:32 |
| SpamapS | IMO the secrets rules are really complicated and hard to predict and test. | 17:33 |
| SpamapS | I need some way to play with them. | 17:33 |
| SpamapS | I guess I can play with them on my staging cluster. | 17:33 |
| mordred | corvus: ^^ am I saying wrong things to SpamapS here? | 17:33 |
| *** maho has left #zuul | 17:36 | |
| pabelanger | secrets are tide to playbooks, so if you have a secret in project-config, only playbooks in project-config have access. You can parent to a job with secrets, which we do today | 17:37 |
| SpamapS | Yes but that's a trusted repo | 17:38 |
| *** markush1 has joined #zuul | 17:39 | |
| *** markush1 has left #zuul | 17:39 | |
| SpamapS | the question is, if I put a job+playbook in my untrusted repo of shared jobs (godaddy-zuul-jobs we call it), will that job work with that secret when I run the job in other repos. | 17:39 |
| SpamapS | My experience was no | 17:39 |
| SpamapS | the variable wasn't set when the playbook ran | 17:39 |
| SpamapS | but again, I may have missed something simple. | 17:40 |
| corvus | SpamapS: i think that should have worked, as long as the playbook was in godaddy-zuul-jobs | 17:41 |
| SpamapS | yeah it was, so probably playbook error | 17:41 |
| SpamapS | it's really hard to test these things because they can't be in check. | 17:42 |
| corvus | SpamapS: you could allow them to be in check temporarily if you're confident you won't leak the secret | 17:42 |
| SpamapS | so now I'm done migrating the job to project-config :-P | 17:42 |
| corvus | that might be a good way of doing things when setting up a new job... especially if you can start with a bogus secret, or it's easy to change the secret when you go into prod | 17:43 |
| *** sshnaidm is now known as sshnaidm|off | 17:49 | |
| *** sshnaidm|off has quit IRC | 17:56 | |
| SpamapS | corvus: yeah I may have a strategy for that actually | 18:05 |
| *** dkranz has joined #zuul | 18:11 | |
| *** jpena is now known as jpena|off | 18:14 | |
| *** persia has quit IRC | 18:39 | |
| *** dkranz has quit IRC | 18:40 | |
| *** harlowja has quit IRC | 19:00 | |
| openstackgerrit | Merged openstack-infra/nodepool master: Clean up old images for testing https://review.openstack.org/544582 | 19:16 |
| openstackgerrit | Merged openstack-infra/nodepool master: Switch to fedora-27 for nodepool dsvm https://review.openstack.org/544583 | 19:16 |
| *** harlowja has joined #zuul | 19:35 | |
| *** harlowja_ has joined #zuul | 19:37 | |
| *** harlowja has quit IRC | 19:39 | |
| *** hasharAway is now known as hashar | 19:59 | |
| *** sshnaidm|off has joined #zuul | 20:08 | |
| openstackgerrit | Paul Belanger proposed openstack-infra/nodepool master: Take 2: Add functional test for key-name and glean https://review.openstack.org/545113 | 20:46 |
| SpamapS | hrm, this is awkward to handle | 20:54 |
| SpamapS | so I have a bunch of stuff waiting on a cloud that is hurting and not finishing building instances | 20:54 |
| SpamapS | I've subsequently turned the max-servers for that cloud to 0 | 20:54 |
| SpamapS | but.. the manager for it still has a bunch of open requests. | 20:55 |
| SpamapS | and rechecks don't seem to make new ones. | 20:55 |
| SpamapS | is there a way to poke nodepool launcher to make it give up that request? | 20:55 |
| SpamapS | http://paste.openstack.org/show/673123/ | 20:56 |
| clarkb | off the top of my head if the cloud responds in error that will happen | 20:58 |
| clarkb | you might be able to nuke the tcp connection | 20:58 |
| clarkb | that will happen meaning request will go back into request list and other provider will pick it up | 20:58 |
| pabelanger | I think you can also add api-timeout to clouds.yaml | 21:01 |
| mordred | api-timeout should only affect the tcp connection layer - it won't likely affect the wait timeout for the ready poll | 21:02 |
| mordred | that's a different setting... | 21:02 |
| mordred | that defaults to timeout=3600 in nodepool ... which should be an hour | 21:03 |
| pabelanger | ++ | 21:04 |
| mordred | and is in turn managed by 'launch-timeout' | 21:04 |
| corvus | SpamapS: if you restart the launcher, it will drop its lock on the requests and something else can pick them up (possibly the same launcher restarted) | 21:04 |
| SpamapS | corvus: ah, I didn't try that | 21:06 |
| *** elyezer has quit IRC | 21:06 | |
| openstackgerrit | Merged openstack-infra/zuul master: Set remote url on every getRepo in merger https://review.openstack.org/535716 | 21:08 |
| corvus | pabelanger: you may be interested in https://review.openstack.org/544562 | 21:09 |
| pabelanger | corvus: cool, will look shortly | 21:10 |
| *** elyezer has joined #zuul | 21:10 | |
| SpamapS | corvus: thanks, that was the magic sauce | 21:23 |
| *** openstack has joined #zuul | 21:47 | |
| *** ChanServ sets mode: +o openstack | 21:47 | |
| corvus | andreaf: i'd like either SpamapS or tobiash to look at it; if one of them can do that today or tomorrow, we can probably restart openstack's instance with that this weekend for use next week | 21:47 |
| pabelanger | I think we just need to restart executors once we land it | 21:47 |
| andreaf | corvus, pabelanger cool thanks | 21:48 |
| corvus | pabelanger: that one requires a little bit of care, it needs a scheduler restart first, then executors | 21:48 |
| corvus | (the executors are going to start assuming some new fields are always present) | 21:49 |
| pabelanger | Oh, right. Changes to config, of course scheduler | 21:49 |
| corvus | it's okay to run new scheduler/old executors with that, but not the other way around | 21:49 |
| openstackgerrit | Andrea Frittoli proposed openstack-infra/zuul-jobs master: Improve file renaming in stage-output https://review.openstack.org/544606 | 21:53 |
| openstackgerrit | Paul Belanger proposed openstack-infra/nodepool master: Take 2: Add functional test for key-name and glean https://review.openstack.org/545113 | 21:53 |
| dmsimard | tobiash: was the concept of executor zones documented somewhere (in a story?) | 22:07 |
| tobiash | dmsimard: I'm not aware of a documentation of that | 22:08 |
| dmsimard | I thought there was *something* somewhere, whether a patch or a story but can't find anything :/ | 22:08 |
| tobiash | I only know about some discussions here about that | 22:08 |
| tobiash | but not when they happened | 22:09 |
| pabelanger | https://storyboard.openstack.org/#!/story/2001125 | 22:12 |
| pabelanger | dmsimard: tobiash: ^ | 22:12 |
| dmsimard | T_T | 22:13 |
| dmsimard | pabelanger: thanks <3 | 22:13 |
| tobiash | corvus: hostvars lgtm | 22:13 |
| dmsimard | Any idea why Zuul v2 failure-pattern/success-pattern would not work? Even url_pattern in zuul.conf doesn't seem to work. Always defaults to the build URL. | 22:26 |
| *** elyezer has quit IRC | 22:26 | |
| dmsimard | I've done something like this: https://github.com/openstack-infra/project-config/blob/8643da6c1c246249d694fb7f7c24d51a7449df94/zuul/layout.yaml#L2713-L2714 | 22:27 |
| openstackgerrit | Merged openstack-infra/zuul master: Fix branch deletion after failed reconfig https://review.openstack.org/544716 | 22:27 |
| *** elyezer has joined #zuul | 22:28 | |
| *** dkranz has joined #zuul | 22:29 | |
| * dmsimard can't wait to migrate to zuul v3 | 22:29 | |
| corvus | dmsimard: i think if that hit a format string error, it would silently fail and use the default | 22:30 |
| corvus | dmsimard: so in that example, maybe LOG_PATH wasn't defined | 22:31 |
| dmsimard | corvus: the included python functions file runs before that is loaded, right ? So https://github.com/openstack-infra/project-config/blob/8643da6c1c246249d694fb7f7c24d51a7449df94/zuul/openstack_functions.py | 22:32 |
| corvus | dmsimard: the functions are run if the job specifies (we had a metajob which ran set_log_url on every job) before that's used, so we were guaranteed to always have LOG_PATH set | 22:35 |
| corvus | the metajob: https://github.com/openstack-infra/project-config/blob/8643da6c1c246249d694fb7f7c24d51a7449df94/zuul/layout.yaml#L1105 | 22:36 |
| dmsimard | corvus: ah, yeah, we have something like that too. | 22:36 |
| dmsimard | ok, I'll keep digging | 22:37 |
| corvus | so consider whether it's possible that's not running, or throwing an exception or something | 22:37 |
| *** threestrands has joined #zuul | 22:43 | |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Make NodeSetParser a regular class https://review.openstack.org/545148 | 22:51 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Make SecretParser a regular class https://review.openstack.org/545149 | 22:51 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Make JobParser a regular class https://review.openstack.org/545150 | 22:51 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Make PipelineParser a regular class https://review.openstack.org/545151 | 22:51 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Make SemaphoreParser a regular class https://review.openstack.org/545152 | 22:51 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Make TenantParser a regular class https://review.openstack.org/545153 | 22:51 |
| *** hashar has quit IRC | 22:52 | |
| mordred | corvus: ooh! | 23:09 |
| corvus | there's a long road ahead, but i see a path from here to a comprehensible configloader and on to memory reduction | 23:11 |
| clarkb | is the issue that addresses being able to have class objects that can be garbage collected? | 23:15 |
| corvus | clarkb: no, it's that i can't understand what the code is doing because it's a rat's nest of function arguments :) | 23:16 |
| corvus | the approach did not scale, and changes to the system take exponentially longer to think about as it has grown | 23:17 |
| corvus | (i'm not done with refactoring yet -- i wouldn't bother if those changes were all i wanted to do) | 23:18 |
| clarkb | gotcha so more mental modelling than direct memory savings to start | 23:18 |
| corvus | yep. because to fix the memory usage, we need to change how things are resolved and cached. and that's really complicated. | 23:19 |
| openstackgerrit | Ian Wienand proposed openstack-infra/nodepool master: Add legacy-copy-jenkins-scripts https://review.openstack.org/545158 | 23:28 |
| openstackgerrit | Merged openstack-infra/nodepool master: Take 2: Add functional test for key-name and glean https://review.openstack.org/545113 | 23:39 |
| clarkb | corvus: why does nodeset parser need a tenant and layout in https://review.openstack.org/#/c/545148/1/zuul/configloader.py ? | 23:40 |
| corvus | clarkb: it doesn't. but a lot do, so i'm toying with the idea of just standardizing on that. | 23:41 |
| openstackgerrit | Ian Wienand proposed openstack-infra/zuul master: Use pip3 in nodepool-integration-setup.sh https://review.openstack.org/545163 | 23:51 |
| openstackgerrit | Ian Wienand proposed openstack-infra/nodepool master: Add legacy-copy-jenkins-scripts https://review.openstack.org/545158 | 23:51 |
| *** rlandy|rover is now known as rlandy|rover|bia | 23:53 | |
| *** rlandy|rover|bia is now known as rlandy|rover|bbl | 23:53 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!