| *** rlandy|bbl is now known as rlandy | 02:23 | |
| *** rlandy has quit IRC | 03:42 | |
| *** rlandy has joined #zuul | 03:50 | |
| *** rlandy has quit IRC | 03:53 | |
| openstackgerrit | Simon Westphahl proposed openstack-infra/nodepool master: Cleanup of leaked resource for static driver https://review.openstack.org/600084 | 05:20 |
|---|---|---|
| openstackgerrit | Simon Westphahl proposed openstack-infra/nodepool master: Implement liveness check for static nodes https://review.openstack.org/601513 | 05:20 |
| SpamapS | hrmph. One thing not fun: debugging a new post job. :-P | 06:15 |
| SpamapS | trying to re-create the fail scenario is.. a challenge | 06:16 |
| openstackgerrit | Simon Westphahl proposed openstack-infra/zuul master: wip: Use merger to get list of files for pull-request https://review.openstack.org/603287 | 06:25 |
| SpamapS | hrm.. nodepool needs a forget command or something | 06:46 |
| SpamapS | aws driver is having some trouble handling "the node is already gone" | 06:46 |
| tristanC | SpamapS: nodepool is indeed pretty rigid when things doesn't go as expected :) maybe the driver is simply missing a try/expect ? | 07:03 |
| *** bhavikdbavishi has joined #zuul | 07:10 | |
| *** hashar has joined #zuul | 07:14 | |
| SpamapS | tristanC: correct | 07:18 |
| SpamapS | I am trying to test that now.. | 07:18 |
| SpamapS | but pbrx is having trouble | 07:18 |
| SpamapS | so.. time for bed | 07:18 |
| openstackgerrit | Simon Westphahl proposed openstack-infra/zuul master: wip: Use merger to get list of files for pull-request https://review.openstack.org/603287 | 07:49 |
| *** jpena|off is now known as jpena | 07:50 | |
| *** persia has quit IRC | 08:06 | |
| *** persia has joined #zuul | 08:07 | |
| *** bhavikdbavishi has quit IRC | 10:23 | |
| *** jpena is now known as jpena|lunch | 11:38 | |
| *** Shrews has quit IRC | 12:00 | |
| *** Shrews has joined #zuul | 12:02 | |
| *** bhavikdbavishi has joined #zuul | 12:30 | |
| *** chkumar|ruck is now known as chkumar|off | 12:32 | |
| *** jpena|lunch is now known as jpena | 12:38 | |
| *** rlandy has joined #zuul | 12:42 | |
| *** samccann has joined #zuul | 13:01 | |
| *** bhavikdbavishi has quit IRC | 13:21 | |
| *** Shrews has quit IRC | 13:27 | |
| *** Shrews has joined #zuul | 13:32 | |
| *** annabelleB has joined #zuul | 14:07 | |
| dmsimard | SpamapS: you can enqueue post jobs manually fwiw | 14:14 |
| *** annabelleB has quit IRC | 14:22 | |
| *** annabelleB has joined #zuul | 14:28 | |
| annabelleB | hey Zuul friends—the Kata folks were looking to chat about their Zuul migration during their weekly meeting. The next one is September 24 at 1500 UTC. Does that work for some of you to join? (If not we can find an outside time) | 14:32 |
| clarkb | annabelleB: would it be more appropriate to talk with Zuul devs or the OpenStack Infra team running the service? Maybe both depending on the specific agenda items. Is that posted anywhere? | 14:49 |
| annabelleB | clarkb: good question. I think the agenda is some of the challenges they’ve had with the transition, similar to that github issue you’ve been on | 14:54 |
| clarkb | annabelleB: I would say 99% of that is an unwillingness to reach out directly to the team running the service | 14:55 |
| clarkb | and I think in the case where that hapepned (beacused I was on vacation) the feedback was ignored? | 14:56 |
| annabelleB | clarkb: would you prefer that I chat with them about “here’s what to do when you’re having problems” vs this sort of, cross-team reset chat? | 14:59 |
| clarkb | I think a reset chat can be useful. I was mostly trying to point out that the problems seem to be communication based and being asked to join a synchronous meeting in kata land seems to perpetuate the communication issues. If kata has problems with zuul or the service the right way to communicate that is to talk to zuul and/or openstack-infra for which we have several venues to do that | 15:01 |
| annabelleB | clarkb: do you all have a weekly meeting of some sort? kata can come your way :) | 15:02 |
| clarkb | annabelleB: for this particular case the openstack-infra team is probably more relevant than the zuul team. The infra team is running the service that kata is trying to consume. Openstack infra can be found on IRC in #openstack-infra, on openstack-infra@lists.openstack.org for asynchronous comms, and has weekly meetings in #openstack-meeting at 1900UTC Tuesdays (currently noon pacific) | 15:04 |
| clarkb | For zuul software concerns there is this channel and zuul-discuss@lists.zuul-ci.org | 15:04 |
| pabelanger | for context, it took the rdoproject about 1 month to migrate from jenkins to zuul. About 95% of the jobs we used a modified version of zuul-migrate, the last 5% we had to rewrite jobs as native zuulv3 jobs. The hardest part by far is crafting the base-minimal / base jobs. | 15:11 |
| clarkb | pabelanger: in this case I think the issues are less about converting existing jobs (we've largely done that piece where possible I think) and more about understanding what zuul does (eg you don't have to merge things yourself) and who to communicate to when a cloud loses some functionality | 15:13 |
| pabelanger | yah, hardest part for me so far, is github workflow. Getting labels right, right branch protections, and not clicking merge buttong in web | 15:14 |
| pabelanger | I haven't looked yet to see what can be hidden from users, aside removing admin access | 15:14 |
| pabelanger | but, I do miss gerrit :) | 15:15 |
| SpamapS | dmsimard: I feel like it may be a good idea to start thinking about a public API for enqueueing and auto-holding.. I'd love to have a "retry" button for instance.. and a "retry-and-hold" | 15:41 |
| tristanC | SpamapS: mhu proposed a spec exactly for that: https://review.openstack.org/562321 | 15:45 |
| SpamapS | awesome | 15:45 |
| SpamapS | because I'm just pushing empty commits right now... because my zuul scheduler runs in a k8s pod that is not `kubectl exec`able since it is hidden deep in the bowels of a private EC2 subnet. | 15:45 |
| SpamapS | so cmdline stuff makes very little sense. | 15:46 |
| mhu | tristanC, SpamapS there's even more! https://review.openstack.org/#/c/576907/ | 15:46 |
| mhu | the spec might actually need to be refreshed in regard of my experiments | 15:47 |
| mhu | yeah it definitely does | 15:47 |
| mhu | jpena, the TC call is over if you wanna join over mumble | 15:54 |
| SpamapS | mhu: this is cool. Adding some comments on 562321. | 15:57 |
| mhu | SpamapS, thanks, I'll rework it tonight | 15:58 |
| *** hashar is now known as hasharAway | 16:00 | |
| openstackgerrit | David Shrewsbury proposed openstack-infra/nodepool master: Use zk connection passed to OpenStack driver https://review.openstack.org/603428 | 16:04 |
| openstackgerrit | David Shrewsbury proposed openstack-infra/nodepool master: Move OpenStack leak code into driver https://review.openstack.org/603429 | 16:04 |
| clarkb | annabelleB: fwiw the infra meeting agenda is at https://wiki.openstack.org/wiki/Meetings/InfraTeamMeeting#Agenda_for_next_meeting I'm happy to add an item for Kata if they want to/can show up today at that venue. Otherwise simply sending an email to the mailing list might get the ball rolling more quickly then we can followup next week in the kata meeting? | 16:07 |
| openstackgerrit | David Shrewsbury proposed openstack-infra/nodepool master: Move OpenStack leak code into driver https://review.openstack.org/603429 | 16:07 |
| annabelleB | clarkb: I’ll ask if they can join today, otherwise yes, I’ll direct them to the ML (which ml do you have in mind—or so many :) ) | 16:08 |
| clarkb | annabelleB: openstack-infra@lists.openstack.org is likely most appropriate for this topic | 16:08 |
| clarkb | thats the talk directly to the people running the infrastructure services mailing list | 16:08 |
| SpamapS | mhu: hah, I should have read 576907 first.. I think JWT is probably the thing to go with here. | 16:11 |
| *** bhavikdbavishi has joined #zuul | 16:12 | |
| SpamapS | mhu: but I really am not a fan of offloading auth to the webserver. | 16:13 |
| *** rlandy is now known as rlandy|brb | 16:14 | |
| SpamapS | It's an old thought that you can just rely on Apache modules for auth and feed data to your webapps. This entangles the webapp in a complex dependency web that is unnecessary. Python is perfectly capable of fetching jwt's. | 16:14 |
| mhu | SpamapS, does that mean we should include user and auth management within zuul? | 16:15 |
| SpamapS | mhu: no, we should just move the machinery for grabbing your tokens from "whatever you run zuul behind" to Cherrypy. | 16:16 |
| SpamapS | Zuul needs to be "batteries included" | 16:18 |
| clarkb | SpamapS: tristanC iirc the nodepool driver is expected to handle the exceptional cases like I tried to delete a resource but it didn't exist. In that case you remove it from zk and moev on | 16:23 |
| SpamapS | clarkb: yeah I have a patch that should work | 16:23 |
| SpamapS | I'm just not sure it's great.. because boto3 is giving a really *generic* error. | 16:23 |
| SpamapS | Could have been any reason | 16:23 |
| SpamapS | which is weird | 16:23 |
| SpamapS | So in the exception handler I need to do a describe and that gives a better error. | 16:24 |
| SpamapS | And I just now fixed pbrx to work (PBKAC) so I'll roll that out to my cluster and see if the two I have stuck disappear. | 16:24 |
| SpamapS | https://github.com/boto/boto3/issues/1057 .... joy | 16:29 |
| clarkb | SpamapS: it is reassuring to learn that openstack isn't alone in having these problems :) | 16:30 |
| *** annabelleB has quit IRC | 16:38 | |
| *** rlandy|brb is now known as rlandy | 16:38 | |
| *** annabelleB has joined #zuul | 16:40 | |
| *** jpena is now known as jpena|off | 17:02 | |
| *** electrofelix has quit IRC | 17:45 | |
| *** bhavikdbavishi has quit IRC | 18:37 | |
| *** annabelleB has quit IRC | 18:56 | |
| *** hasharAway is now known as hashar | 18:57 | |
| *** annabelleB has joined #zuul | 19:00 | |
| tobiash | SpamapS: sounds like you have fun improving the aws driver ;) | 19:18 |
| *** gouthamr has quit IRC | 20:00 | |
| *** dmellado has quit IRC | 20:01 | |
| *** gouthamr has joined #zuul | 20:03 | |
| *** samccann has quit IRC | 20:09 | |
| *** hashar has quit IRC | 20:47 | |
| *** hashar has joined #zuul | 20:54 | |
| *** jiapei has joined #zuul | 20:54 | |
| *** hashar has quit IRC | 21:03 | |
| SpamapS | tobiash: yeah, it's working alright | 21:13 |
| *** gouthamr_ has joined #zuul | 21:18 | |
| *** dmellado has joined #zuul | 21:20 | |
| *** gouthamr has quit IRC | 21:26 | |
| *** rlandy has quit IRC | 21:31 | |
| dmsimard | SpamapS: yeah the ec2 driver needs work :/ | 21:38 |
| dmsimard | I'll probably be able to revisit but it's on the backburner for now | 21:38 |
| pabelanger | dmsimard: I was thinking of giving it a try later this week. | 21:46 |
| pabelanger | so, might have some feedback / updates | 21:46 |
| openstackgerrit | Sorin Sbarnea proposed openstack-infra/zuul master: Assure that status tooltip is displayed on entire row https://review.openstack.org/603504 | 21:51 |
| SpamapS | dmsimard: it's.. pretty great right now | 21:57 |
| SpamapS | bugs happen | 21:57 |
| pabelanger | SpamapS: did you make process on diskimages? | 21:58 |
| SpamapS | pabelanger: No, I have yet to deploy that code, I've just tested it locally and it *does* call the right EC2 API, but next I need to make sure it actually boots a usable image. | 21:58 |
| clarkb | SpamapS: using dib? | 21:59 |
| SpamapS | yeah | 22:00 |
| SpamapS | It's actually pretty easy to upload a raw image to AWS these days | 22:00 |
| SpamapS | just stick it into S3 and call the import API | 22:00 |
| clarkb | oh you don't have to do the split kernel and image thing? | 22:00 |
| clarkb | thats nice | 22:00 |
| *** sshnaidm|ptg has joined #zuul | 22:00 | |
| SpamapS | correct | 22:00 |
| pabelanger | cool | 22:00 |
| SpamapS | that's old school xen AFAICT | 22:00 |
| pabelanger | yah, happy to test that too | 22:00 |
| *** sshnaidm|ptg is now known as sshnaidm|pto | 22:00 | |
| SpamapS | It's a side project so.. when I get time. | 22:00 |
| *** annabelleB has quit IRC | 22:36 | |
| *** annabelleB has joined #zuul | 22:37 | |
| *** annabelleB has quit IRC | 22:56 | |
| *** sshnaidm|pto has quit IRC | 22:59 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!