*** xinliang has quit IRC | 00:41 | |
*** xinliang has joined #zuul | 00:54 | |
*** harlowja has quit IRC | 02:13 | |
*** rlandy has quit IRC | 02:25 | |
openstackgerrit | Ian Wienand proposed openstack-infra/zuul-jobs master: Add arm64 support to mirror setup https://review.openstack.org/550312 | 02:51 |
---|---|---|
*** dkranz has joined #zuul | 03:29 | |
openstackgerrit | Merged openstack-infra/zuul-jobs master: Handle repos with no tags in version-from-git https://review.openstack.org/550175 | 04:06 |
openstackgerrit | Merged openstack-infra/zuul-jobs master: Add arm64 support to mirror setup https://review.openstack.org/550312 | 05:28 |
*** snapiri has quit IRC | 05:56 | |
*** bhavik has joined #zuul | 05:59 | |
*** bhavik has quit IRC | 06:43 | |
*** swest has quit IRC | 06:45 | |
openstackgerrit | Andreas Jaeger proposed openstack-infra/zuul-jobs master: Fix version-from-git syntax https://review.openstack.org/550357 | 06:52 |
*** threestrands_ has joined #zuul | 07:09 | |
*** threestrands_ has quit IRC | 07:09 | |
*** threestrands_ has joined #zuul | 07:09 | |
*** threestrands has quit IRC | 07:10 | |
*** threestrands_ has quit IRC | 07:13 | |
openstackgerrit | Merged openstack-infra/zuul-jobs master: Fix version-from-git syntax https://review.openstack.org/550357 | 07:20 |
*** hashar has joined #zuul | 07:53 | |
*** jpena|off is now known as jpena | 08:07 | |
openstackgerrit | Andreas Jaeger proposed openstack-infra/zuul-jobs master: Fix condition in version-from-git https://review.openstack.org/550388 | 08:22 |
openstackgerrit | Andreas Jaeger proposed openstack-infra/zuul-jobs master: Fix condition in version-from-git https://review.openstack.org/550388 | 08:25 |
openstackgerrit | Tobias Henkel proposed openstack-infra/zuul master: DNM: run tox-cover job https://review.openstack.org/550392 | 08:32 |
*** threestrands has joined #zuul | 08:32 | |
*** threestrands has quit IRC | 08:32 | |
*** threestrands has joined #zuul | 08:32 | |
openstackgerrit | Tobias Henkel proposed openstack-infra/zuul master: Support pagination in the installation map https://review.openstack.org/550393 | 08:35 |
tobiash | SpamapS: in case you have more than 100 repos in an organization ^ | 08:35 |
*** threestrands has quit IRC | 08:36 | |
tobiash | one of my users added 600 repos to his org and crashed zuul config with that... | 08:36 |
*** abelur has joined #zuul | 08:44 | |
*** electrofelix has joined #zuul | 08:47 | |
openstackgerrit | Merged openstack-infra/zuul-jobs master: Fix condition in version-from-git https://review.openstack.org/550388 | 08:51 |
*** swest has joined #zuul | 08:56 | |
openstackgerrit | Tobias Henkel proposed openstack-infra/zuul master: DNM: run tox-cover job https://review.openstack.org/550392 | 09:01 |
openstackgerrit | Fabien Boucher proposed openstack-infra/zuul master: Tenant config can be read from an external script https://review.openstack.org/535878 | 09:12 |
openstackgerrit | Merged openstack-infra/zuul-jobs master: Revert "Revert "Add zuul.{pipeline,nodepool.provider,executor.hostname} to job header"" https://review.openstack.org/514489 | 09:19 |
openstackgerrit | Merged openstack-infra/zuul-jobs master: Add Ansible version to job header https://review.openstack.org/532304 | 09:19 |
*** swest has quit IRC | 10:07 | |
*** swest has joined #zuul | 10:24 | |
openstackgerrit | Andrea Frittoli proposed openstack-infra/zuul-jobs master: Add known hosts for all users https://review.openstack.org/548642 | 10:27 |
openstackgerrit | Andreas Jaeger proposed openstack-infra/zuul-jobs master: Fix emit-job-header https://review.openstack.org/550425 | 10:28 |
openstackgerrit | Andreas Jaeger proposed openstack-infra/zuul-jobs master: Fix emit-job-header https://review.openstack.org/550425 | 10:53 |
openstackgerrit | Andrea Frittoli proposed openstack-infra/zuul-jobs master: Add known hosts for all users https://review.openstack.org/548642 | 11:03 |
openstackgerrit | Merged openstack-infra/zuul-jobs master: Fix emit-job-header https://review.openstack.org/550425 | 11:21 |
*** jpena is now known as jpena|lunch | 12:03 | |
mordred | tobiash: left a note on 550393 | 12:17 |
mordred | tobiash: (which I only konw because I added support for the same thing to openstacksdk about a month ago) | 12:17 |
tobiash | oh, cool, thx :) | 12:17 |
*** myoung|ruck|off is now known as myoung|ruck | 12:19 | |
*** yolanda has quit IRC | 12:20 | |
*** yolanda has joined #zuul | 12:21 | |
*** myoung|ruck is now known as myoung|ruck|mtg | 12:59 | |
*** jpena|lunch is now known as jpena | 13:05 | |
*** rlandy has joined #zuul | 13:19 | |
*** mickt has joined #zuul | 13:41 | |
mickt | Hi, setting up a zuul v3 system but finding documentation somewhat lacking. Anyone have a link to something useful. Maybe a use case for e.g. openstack-keystone or similar. | 13:45 |
tristanC | mickt: we wrote some practical zuul usage blog post here: https://blogs.rdoproject.org/?s=zuul | 13:47 |
mickt | Cool, I'll take a look. Thanks | 13:47 |
clarkb | mickt: might help if you could expand a bit more on what particular bits you are struggling with. Is it setting up the services or configuring jobs or something else? also not sure what you mean by a use case for keystone? | 13:48 |
clarkb | also the official docs are at https://docs.openstack.org/infra/zuul/ | 13:48 |
mickt | job configuration etc. I'm a newbie to this too, which doesn't help | 13:48 |
clarkb | mickt: the users guide https://docs.openstack.org/infra/zuul/user/index.html in particular tries to walk you through job configuration | 13:50 |
clarkb | mickt: I would probably walk through it step by step. If you plan on running your own zuul then start by getting the services up. Then as you push configs to it they will largely be self testing so you can iterate and see how things fit together | 13:59 |
clarkb | and if you have specific questions (like why doesn't this particular bit of ansible work or what is this zuul service doing) we can help sort that out either here or on the mailing list and update docs as appropriate | 14:00 |
*** myoung|ruck|mtg is now known as myoung|ruck | 14:03 | |
openstackgerrit | Tobias Henkel proposed openstack-infra/zuul master: Support pagination in the installation map https://review.openstack.org/550393 | 14:09 |
tobiash | mordred: ^ | 14:10 |
mordred | tobiash: loks great! | 14:12 |
tobiash | :) | 14:14 |
mickt | So clarkb, I have a zuul-merger and two zuul-scheduler processes running. If I make a change to e.g. openstack-keystone repo and submit for review I would see the tests performed and failed etc.? | 14:15 |
clarkb | does github gracefully give you fewer results if you say ask for 1000 per page? (thinking that would be useful if they lower then limit in the future) | 14:15 |
tobiash | mordred: that was the outcome of yesterdays lunch firefighting ;) | 14:15 |
tobiash | clarkb: yes, you can request 1000 but still get 100 | 14:15 |
clarkb | mickt: at the very least you should see the scheduler attempt to queue up some jobs | 14:16 |
tobiash | my first try was to request 1000 ;) | 14:16 |
clarkb | tobiash: perfect (this way they can change without everyone needing to update zuul) | 14:16 |
tobiash | yes | 14:16 |
mickt | ok, so I need to look into 'gear.GearmanError: Unable to submit job to any connected servers' | 14:24 |
clarkb | mickt: yup. My guess is that you don't have any executors running so no gearman workers are registered | 14:24 |
clarkb | mickt: you can check that by connecting to port 4730 on the scheduler and entering 'status' without the quotes (thats a gearman command to get back status info including registered jobs) | 14:25 |
openstackgerrit | Merged openstack-infra/zuul master: Support pagination in the installation map https://review.openstack.org/550393 | 14:37 |
tobiash | \o/ | 14:48 |
*** electrofelix has quit IRC | 14:51 | |
*** swest has quit IRC | 14:51 | |
openstackgerrit | Monty Taylor proposed openstack-infra/zuul-jobs master: Add nodepool provider back to job header https://review.openstack.org/550493 | 15:01 |
*** mhu has quit IRC | 15:02 | |
*** pabelanger has quit IRC | 15:04 | |
*** rlandy has quit IRC | 15:09 | |
*** rlandy has joined #zuul | 15:09 | |
*** rlandy has quit IRC | 15:10 | |
tobiash | mordred: I've a comment on ^ | 15:15 |
tobiash | would you agree? | 15:15 |
*** pabelanger has joined #zuul | 15:17 | |
openstackgerrit | Paul Belanger proposed openstack-infra/zuul-jobs master: Fix lint issue with emit-job-header role https://review.openstack.org/550501 | 15:20 |
tobiash | pabelanger: what is the correct way to remove an image? | 15:31 |
pabelanger | tobiash: remove from nodepool.yaml? | 15:31 |
tobiash | I set the label to min-ready 0, then removed the label | 15:32 |
tobiash | yes | 15:32 |
tobiash | now I thought, delete the images | 15:32 |
clarkb | you remove the image from all the pools first | 15:32 |
tobiash | but they are reuploaded | 15:32 |
pabelanger | yah, you need to remove image from providers.pool | 15:32 |
pabelanger | to stop uploads | 15:32 |
tobiash | does it know how to delete it then? | 15:32 |
clarkb | that means don't upload this image to the pool | 15:32 |
clarkb | tobiash: yes it will delete it too iirc | 15:32 |
pabelanger | yup | 15:32 |
tobiash | ok, trying | 15:32 |
pabelanger | you might need to set max-servers: -1 for that image, to delete ready nodes | 15:33 |
pabelanger | if you have any | 15:33 |
tobiash | nodes are already gone | 15:33 |
pabelanger | k | 15:33 |
pabelanger | then images: [] in providers.pool should delete them | 15:33 |
pabelanger | Shrews: mind a review adding test coverage in nodepool https://review.openstack.org/540916/ | 15:36 |
tobiash | pabelanger, clarkb: thanks, worked | 15:37 |
pabelanger | yay | 15:37 |
tobiash | and after that remove the diskimage? | 15:37 |
pabelanger | we really should update docs if lacking | 15:37 |
pabelanger | tobiash: nodepool-builder should, if no providers are using it | 15:37 |
tobiash | ah, it's already gone | 15:38 |
pabelanger | woot | 15:38 |
Shrews | pabelanger: sure. i have several nodepool reviews that could use some love too | 15:49 |
Shrews | pabelanger: hrm, i do not think that would pass a recheck. corvus made a change to default min_ready to 0 around that time | 15:52 |
pabelanger | oh, right. | 15:53 |
pabelanger | I can fix that | 15:53 |
Shrews | pabelanger: oh, maybe it would. i was looking at the new config. i think it might still be ok | 15:54 |
Shrews | pabelanger: +3'd | 15:54 |
pabelanger | okay, if it fails, I can update | 15:55 |
Shrews | yep | 15:55 |
clarkb | if it doesn't fail does that imply the test is also broken? | 15:55 |
*** myoung|ruck is now known as myoung|ruck|brb | 15:55 | |
clarkb | (eg not testing what we intend) | 15:55 |
Shrews | clarkb: no. only thing broken are my eyes | 15:56 |
Shrews | btw, the zookeeper retry discussion in Zuul Discuss has been pretty eye opening to bugs in our code. very glad that was started | 15:57 |
Shrews | thanks tristanC! | 15:57 |
openstackgerrit | Merged openstack-infra/zuul-jobs master: Fix lint issue with emit-job-header role https://review.openstack.org/550501 | 15:58 |
pabelanger | yah, I'd like to see us fix the start zuul-sceduler with zookeeper firewalled bug. Where kazoo.client will never retry a connection after a few attempts. | 15:59 |
Shrews | pabelanger: i think we can fix that using part of tristanC's change for the initial connect. it's the code after connect that i have concerns with. | 16:06 |
pabelanger | Shrews: yah, that is what I suspected | 16:07 |
openstackgerrit | Merged openstack-infra/nodepool master: Add unit test for multiple launchers https://review.openstack.org/540916 | 16:12 |
Shrews | Anyone want to show these nodepool reviews some love? https://review.openstack.org/545867 , https://review.openstack.org/546303 , https://review.openstack.org/548376 | 16:13 |
Shrews | 1st two are dead simple | 16:13 |
clarkb | Shrews: probably a good way for me to fight jet lag and get back into the swing of things. I will take a look | 16:13 |
Shrews | clarkb: thx! | 16:14 |
mickt | Should I have a running zuul-executor service or is it run as required? | 16:14 |
clarkb | mickt: you will need a running zuul-executor service | 16:14 |
mickt | ok | 16:14 |
dmsimard | https://landscape.cncf.io/selected=zuul-ci \o/ | 16:15 |
pabelanger | mickt: you likely also want zuul-web, zuul-fingergw. zuul-merger would be optional, but maybe needed as you grow more projects / git operations | 16:17 |
*** myoung|ruck|brb is now known as myoung|ruck | 16:17 | |
mordred | dmsimard: \o/ indeed | 16:20 |
*** rlandy has joined #zuul | 16:23 | |
openstackgerrit | Monty Taylor proposed openstack-infra/zuul-jobs master: Add copy of emit-job-header to test header change https://review.openstack.org/550518 | 16:29 |
openstackgerrit | Monty Taylor proposed openstack-infra/zuul-jobs master: Add nodepool provider back to job header https://review.openstack.org/550493 | 16:30 |
openstackgerrit | Monty Taylor proposed openstack-infra/zuul-jobs master: Remove test-emit-job-header https://review.openstack.org/550521 | 16:31 |
*** tobasco is now known as tobasco|away | 16:35 | |
*** weshay is now known as weshay|PTO | 16:38 | |
openstackgerrit | Merged openstack-infra/nodepool master: Simplify launcher cleanup worker https://review.openstack.org/545867 | 16:38 |
openstackgerrit | Merged openstack-infra/nodepool master: Add additional builder debug logging https://review.openstack.org/546303 | 16:47 |
clarkb | Shrews: for 548376 a couple of questions. Why does the label list need to be sorted? (you chose not to use a set because you can't sort them but its not clear to me why sorting is necessary). And if A provider is removed from the launcher config we don't seem to remove the registration from zookeeper? we will remove it when the launcher loses its connection to zk due to the ephemeral = true though | 16:59 |
Shrews | clarkb: good questions. I think I wanted them always sorted for comparison purposes, mostly. Though I'm sure we'd also want them sorted during display too. For the 2nd question, that's a good point. When we register, we aren't considering the labels from the provider sections, only what labels are defined in the outer 'labels' config. I suppose it's possible to have a label defined there that actually isn't | 17:05 |
Shrews | supplied by a provider? | 17:05 |
clarkb | Shrews: ya I think that is my concern that we'll tell zuul that we support a label but not actually be able to boot one | 17:05 |
clarkb | right now its "ok" because we don't really have a contract with zuul on what is available it just asks for stuff and gets what it can | 17:06 |
Shrews | clarkb: let me put some thought into that bit. i want to see if nodepool will reject a label that isn't used in a provider (don't think it does), and if it doesn't, see if i can only pull the ones that we can really supply | 17:07 |
Shrews | clarkb: feel free to -1 for that if you want | 17:08 |
clarkb | ok will do | 17:08 |
Shrews | clarkb: thanks for reviewing | 17:08 |
mickt | So of all the zuul services, which should be always running? Currently I have scheduler, merger & executor. | 17:24 |
clarkb | mickt: the bare minimum is scheduler and executor I think. The executor will act as its own merge if there is not explicit merger service. And all the web stuff is for ux but shouldn't prevent jobs from running | 17:25 |
tobiash | mickt: if you connect to github also zuul-web is a must | 17:25 |
tobiash | if not it is a *really want have* | 17:25 |
tobiash | merger is optional and only needed for scaling | 17:26 |
pabelanger | zuul-fingergw allows for streaming jobs too | 17:26 |
*** myoung|ruck is now known as myoung|lunch | 17:40 | |
*** jpena is now known as jpena|off | 17:42 | |
*** abadger1999 has left #zuul | 18:39 | |
*** myoung|lunch is now known as myoung|ruck | 18:49 | |
*** harlowja has joined #zuul | 19:26 | |
SpamapS | was zuul-fingergw built into scheduler before? | 19:39 |
SpamapS | I need to rebase my local stack and rework my ansible if so. :-P | 19:39 |
tobiash | SpamapS: no, it never was | 19:40 |
clarkb | SpamapS: I don't think it existed atall instead you had to talk to the executors finger port directly | 19:40 |
clarkb | SpamapS: fingergw basically acts as a multiplexor for finger to the executors and is new | 19:40 |
tobiash | SpamapS: it was already a separate daemon when it was introduced | 19:40 |
clarkb | this allows you to run the executors on non privileged ports too which is nice | 19:41 |
SpamapS | clarkb: OH ok | 19:42 |
SpamapS | I only have one executor. ;) | 19:42 |
tobiash | SpamapS: but the javascript zuul-web stuff has landed so you might want to check if you have to change something for zuul-web deployment | 19:42 |
tobiash | e.g. yarn is a new dependency | 19:43 |
SpamapS | Yeah I'll have to adapt some | 19:49 |
SpamapS | http://paste.openstack.org/show/693716/ | 19:49 |
SpamapS | I think ultimately it gets simpler. :) | 19:49 |
SpamapS | I just wish we had like, a linux distro that wasn't so dang slow *cough*centos7*cough* | 19:50 |
openstackgerrit | David Shrewsbury proposed openstack-infra/nodepool master: Store label info with launcher registration https://review.openstack.org/548376 | 19:52 |
Shrews | clarkb: i think that ^ will get us what we want | 19:53 |
clarkb | Shrews: cool I'll rereview now | 19:53 |
Shrews | clarkb: i also changed to using a set() instead of list() for the labels | 19:53 |
clarkb | Shrews: the check for static vs openstack looks like maybe a bug in our driver interface? (probably not something to fix in this cahnge but something to address before we add a bunch of drivers) | 19:54 |
tobiash | SpamapS: with the latest zuul-web changes you probably can just skip almost all of your paste ;) | 19:55 |
Shrews | clarkb: well, not really a bug so much as each can have a unique configuration | 19:55 |
clarkb | Shrews: ya maybe it shouldn't be unique for the label mapping? | 19:55 |
clarkb | or maybe we need a driver function to get the labels from it? | 19:56 |
Shrews | clarkb: or at least an API call we can depend on to return us that info, yeah | 19:56 |
Shrews | tristanC: ^^^ | 19:56 |
Shrews | all new territory :) | 19:56 |
clarkb | as a user it might be nice to have some uniformity to the config but I can already foresee that some drivers will be different enough that configs won't all be so similar. Happy to go either way on that one but a driver api for get labels probably the most flexible option | 19:58 |
tobiash | Shrews: are the labels from the pools really sufficient? | 19:58 |
Shrews | tobiash: are they not? | 19:58 |
tobiash | we also have the globally defined labels | 19:58 |
tobiash | where min-ready is defined | 19:58 |
tobiash | if a label is missing there but defined in a pool, will that be available? | 19:59 |
clarkb | tobiash: thats a good question | 19:59 |
tobiash | or is that a config error? | 19:59 |
clarkb | I woudl expect that to be a config error | 19:59 |
Shrews | i thought that would be a config error | 19:59 |
Shrews | maybe not? | 19:59 |
clarkb | but I'm not sure that is current behavior | 19:59 |
clarkb | Shrews: you can probably update your test to test that really quickly | 19:59 |
Shrews | i get a KeyError from the fake/openstack driver | 20:01 |
clarkb | cool I think it does the right thing in that situation then | 20:01 |
clarkb | left a comment about the driver api thing, but otherwise +2 | 20:02 |
Shrews | static driver also gives me a KeyError, but I don't think our config validator catches that for either (doesn't for static) | 20:03 |
tobiash | yah, defining a non-existent label in a pool should crash here: http://git.openstack.org/cgit/openstack-infra/nodepool/tree/nodepool/driver/openstack/config.py#n205 | 20:04 |
*** mickt has quit IRC | 20:47 | |
*** hashar has quit IRC | 21:00 | |
Shrews | oh yay, that's a fun error from the launcher change. i wonder if i should just log.error() when an invalid driver is encountered, rather than raise an exception? | 21:40 |
clarkb | Shrews: that might make it easier for tristanc to carry patches locally too until we merge more drivers (and fix the config loading of labels thing) | 21:40 |
Shrews | clarkb: tobiash: thoughts on that? for reference, http://logs.openstack.org/76/548376/2/check/tox-py35/e68a44b/job-output.txt.gz#_2018-03-07_20_01_55_657770 | 21:40 |
Shrews | clarkb: yeah | 21:41 |
openstackgerrit | David Shrewsbury proposed openstack-infra/nodepool master: Store label info with launcher registration https://review.openstack.org/548376 | 21:42 |
Shrews | that failed test passes locally with that change | 21:43 |
tobiash | Shrews: lgtm | 21:46 |
tobiash | mordred: I have totally no idea of ejs, but could it be that the '-' breaks this? | 21:54 |
tobiash | <td>{{ tenant.queue-size }}</td> | 21:54 |
tobiash | I found no other difference to the number of projects | 21:54 |
clarkb | Shrews: one last thought on that is I think this updates the zk api between nodepool and zuul so we should probably consider writing that down somewhere. But probably not in this change? I think current docs of that are the zuul v3 spec? | 21:55 |
clarkb | Shrews: I've approved the change | 21:56 |
Shrews | clarkb: this doesn't affect anything between zuul and nodepool, but yes, the only docs we have of that are the v3 spec | 21:57 |
Shrews | clarkb: this will be used by the nodepool webapp for /label_list | 21:57 |
clarkb | Shrews: oh I was expecting this to be used by zuul to not queue jobs indefinitely if no providers provide the label | 21:57 |
Shrews | something more formal is definitely called for, though. along with a proper library | 21:57 |
Shrews | clarkb: zuul *could* use it | 21:58 |
mordred | tobiash: oh - maybe so? I was actually planning some follow-up work to switch those files away from using the ejs processor since we don't actually use ejs at all | 21:58 |
Shrews | but doubt it will | 21:58 |
mordred | tobiash: I'll get some patches up for that as soon as I'm done with my current thing | 21:58 |
mordred | tobiash: (those should be angular substitutions rather than ejs substitutions, but the fact that the files are in .ejs files makes the situation confusion I believe | 21:59 |
tobiash | mordred: I'll just try queue instead of queue-size | 22:00 |
mordred | cool | 22:00 |
tobiash | but will try that tomorrow | 22:02 |
tobiash | too late here.. | 22:02 |
tobiash | cya | 22:02 |
*** openstackstatus has quit IRC | 22:13 | |
*** openstack has joined #zuul | 22:17 | |
*** ChanServ sets mode: +o openstack | 22:17 | |
*** openstackstatus has quit IRC | 22:42 | |
*** openstack has joined #zuul | 22:46 | |
*** ChanServ sets mode: +o openstack | 22:46 | |
openstackgerrit | Monty Taylor proposed openstack-infra/zuul master: Actually honor the static_path config value https://review.openstack.org/550636 | 22:53 |
*** myoung|ruck is now known as myoung|off | 23:11 | |
*** threestrands has joined #zuul | 23:12 | |
openstackgerrit | Merged openstack-infra/zuul master: Actually honor the static_path config value https://review.openstack.org/550636 | 23:23 |
*** threestrands_ has joined #zuul | 23:34 | |
*** threestrands_ has quit IRC | 23:34 | |
*** threestrands_ has joined #zuul | 23:34 | |
*** threestrands_ has quit IRC | 23:35 | |
*** threestrands_ has joined #zuul | 23:36 | |
*** threestrands has quit IRC | 23:36 | |
*** rlandy is now known as rlandy|bbl | 23:58 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!