| *** rlandy has quit IRC | 00:47 | |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul master: Add data_files entry for zuul/web/static https://review.openstack.org/631494 | 01:25 |
|---|---|---|
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul master: Add data_files entry for zuul/web/static https://review.openstack.org/631494 | 01:47 |
| pabelanger | mordred: I can't see to figure out why pbrx jobs are failing on ^, do you mind looking when you have free time | 01:48 |
| SpamapS | fungi: I tend to see equivs as "I give up packaging stuff". ;) | 01:50 |
| SpamapS | but this case seems like the corner it was ment to handle | 01:50 |
| fungi | well, yes. i mean it's a handy workaround of course | 01:57 |
| *** irclogbot_3 has quit IRC | 03:16 | |
| *** irclogbot_3 has joined #zuul | 03:24 | |
| *** bhavikdbavishi has joined #zuul | 04:56 | |
| *** saneax has joined #zuul | 05:34 | |
| *** badboy has joined #zuul | 06:14 | |
| *** spsurya has joined #zuul | 06:21 | |
| *** quiquell|off is now known as quiquell | 07:12 | |
| quiquell | AJaeger: o/ | 07:19 |
| quiquell | https://review.openstack.org/#/c/623294 | 07:20 |
| *** saneax has quit IRC | 07:20 | |
| quiquell | Is this all good now? | 07:20 |
| quiquell | mordred: ∆ | 07:21 |
| *** saneax has joined #zuul | 07:27 | |
| *** saneax has quit IRC | 07:30 | |
| *** pcaruana has joined #zuul | 07:41 | |
| *** hashar has joined #zuul | 08:21 | |
| *** avass has joined #zuul | 08:27 | |
| *** electrofelix has joined #zuul | 08:56 | |
| *** jpena|off is now known as jpena | 08:56 | |
| *** saneax has joined #zuul | 09:20 | |
| *** saneax has quit IRC | 09:31 | |
| *** saneax has joined #zuul | 09:31 | |
| *** sanjayu_ has joined #zuul | 09:34 | |
| *** saneax has quit IRC | 09:37 | |
| *** bhavikdbavishi has quit IRC | 09:51 | |
| *** sanjayu_ has quit IRC | 10:05 | |
| *** sanjayu_ has joined #zuul | 10:05 | |
| *** hashar has quit IRC | 10:52 | |
| tobias-urdin | could somebody please review https://review.openstack.org/#/c/632163/ :) | 10:53 |
| tobias-urdin | so we can continue testing to get it working | 10:53 |
| *** bhavikdbavishi has joined #zuul | 10:56 | |
| tobiash | tobias-urdin: lgtm, is this role already in use somewhere? | 11:32 |
| *** quiquell is now known as quiquell|lunch | 11:50 | |
| *** dkehn has quit IRC | 11:51 | |
| avass | what does the zuul-daemon do? | 12:02 |
| avass | I mean the zuul_console daemon on the hosts | 12:03 |
| *** bhavikdbavishi has quit IRC | 12:07 | |
| *** hjensas has quit IRC | 12:21 | |
| *** bhavikdbavishi has joined #zuul | 12:36 | |
| *** jpena is now known as jpena|lunch | 12:43 | |
| openstackgerrit | Simon Westphahl proposed openstack-infra/zuul master: List changed files for all commits between refs https://review.openstack.org/631797 | 12:45 |
| *** quiquell|lunch is now known as quiquell | 12:54 | |
| *** hashar has joined #zuul | 12:55 | |
| tobias-urdin | tobiash: well yes and no, we are trying to get it working in openstack ci but not working just yet, hopefully after that patch :) | 13:07 |
| *** bhavikdbavishi has quit IRC | 13:17 | |
| *** bhavikdbavishi has joined #zuul | 13:19 | |
| *** bhavikdbavishi has quit IRC | 13:28 | |
| openstackgerrit | Matthieu Huin proposed openstack-infra/zuul-jobs master: install-nodejs: add support for RPM-based OSes https://review.openstack.org/631049 | 13:32 |
| *** rlandy has joined #zuul | 13:32 | |
| *** AJaeger has quit IRC | 13:33 | |
| mordred | avass: that's what currently streams the console logs back to the executor so that they can be shown in the web dashboard and finger gateway | 13:41 |
| mordred | avass: there is half-finished work to rework how that is done and remove the need for the zuul_console daemon | 13:42 |
| *** jpena|lunch is now known as jpena | 13:47 | |
| avass | mordred: ah, just what i was looking for. I'm guessing there's no windows version of it, right? | 13:48 |
| mordred | avass: no, unfortunately not. I think the rework to replace zuul_console will work much better on windows | 13:49 |
| avass | mordred: alright | 13:49 |
| mordred | sorry bout that | 13:49 |
| *** bhavikdbavishi has joined #zuul | 13:51 | |
| *** badboy has quit IRC | 13:51 | |
| *** gtema has joined #zuul | 13:55 | |
| *** AJaeger has joined #zuul | 14:05 | |
| *** gtema has quit IRC | 14:14 | |
| *** gtema has joined #zuul | 14:14 | |
| *** sanjayu_ has quit IRC | 14:19 | |
| *** sanjayu_ has joined #zuul | 14:20 | |
| *** sanjayu_ has quit IRC | 14:26 | |
| quiquell | mordred: Do zuul have any limitation on the kind of keys you have to set at sshkey at gerrit connection | 14:32 |
| quiquell | mordred: also can you use anonymous user there ? | 14:32 |
| *** saneax has joined #zuul | 14:46 | |
| *** avass has quit IRC | 15:02 | |
| *** hashar has quit IRC | 15:22 | |
| *** avass has joined #zuul | 15:24 | |
| *** quiquell is now known as quiquell|off | 15:25 | |
| openstackgerrit | Monty Taylor proposed openstack-infra/nodepool master: Build images with Dockerfile instead of pbrx https://review.openstack.org/632187 | 15:38 |
| *** gtema has quit IRC | 15:42 | |
| pabelanger | mordred: tobiash: do you have any thoughts on what is going on with https://review.openstack.org/631494/ and pbrx jobs? I am modifying the web/static folder location | 15:48 |
| *** openstackgerrit has quit IRC | 15:51 | |
| mordred | pabelanger: no - I haven't dug in to it too much. I'm not 100% sure that's going to work though - I thought with wheel-based installs python stopped supporting installing things into arbitrary locations | 15:59 |
| mordred | pabelanger: however, it's also possible there is an issue because zuul/web/static/* is generated | 15:59 |
| mordred | pabelanger: I'm not 100% sure though - the logs are a bit weird - I'll need to dig in a bit more, but I've got a phone call I need to jump on rightnow | 16:00 |
| pabelanger | mordred: okay thanks, I'll keep digging too | 16:01 |
| dmsimard | vars in zuul jobs declaration are inventory vars, right ? not extra-vars ? | 16:09 |
| *** openstackgerrit has joined #zuul | 16:10 | |
| openstackgerrit | Gabriele Cerami proposed openstack-infra/zuul master: Scheduler: make autohold hold_list configurable https://review.openstack.org/632498 | 16:10 |
| openstackgerrit | Sagi Shnaidman proposed openstack-infra/nodepool master: Support userdata for instances in openstack https://review.openstack.org/630649 | 16:24 |
| *** ianychoi has joined #zuul | 16:26 | |
| Shrews | mordred: where is the change to publish opendevorg/python-builder? | 16:33 |
| Shrews | err, promote rather | 16:34 |
| Shrews | ah, found it | 16:35 |
| Shrews | but it haz sads | 16:35 |
| *** avass has quit IRC | 16:46 | |
| *** hashar has joined #zuul | 16:59 | |
| *** panda is now known as panda|off | 17:02 | |
| *** pcaruana has quit IRC | 17:02 | |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul-jobs master: Add docker image build jobs https://review.openstack.org/632172 | 17:02 |
| openstackgerrit | Sagi Shnaidman proposed openstack-infra/nodepool master: Support userdata for instances in openstack https://review.openstack.org/630649 | 17:15 |
| tobiash | dmsimard: yes, should be inventory vars | 17:36 |
| openstackgerrit | Tobias Henkel proposed openstack-infra/zuul master: Add spec for multi ansible version support https://review.openstack.org/623927 | 17:43 |
| *** bhavikdbavishi has quit IRC | 17:55 | |
| *** pcaruana has joined #zuul | 18:01 | |
| *** ssbarnea|rover has joined #zuul | 18:02 | |
| *** ssbarnea|bkp2 has quit IRC | 18:04 | |
| *** hashar is now known as hasharDinner | 18:05 | |
| *** jpena is now known as jpena|off | 18:22 | |
| *** pcaruana has quit IRC | 18:23 | |
| SpamapS | is there any progress on a buildset page? | 18:25 |
| SpamapS | The default role that makes zuul log paths is chopping all but the last 2 chars off of PR#'s, so my GitHub status URLs are all wrong once a repo gets past PR#99. | 18:26 |
| SpamapS | and I can't do any transforms in the pipeline config.. so... status URL is officially busted. :-/ | 18:26 |
| SpamapS | Oh hm | 18:27 |
| SpamapS | I guess I can do /builds?change=xxx | 18:27 |
| *** saneax has quit IRC | 18:38 | |
| pabelanger | dmsimard: depends, https://zuul-ci.org/docs/zuul/user/jobs.html#variables | 18:44 |
| *** hasharDinner has quit IRC | 18:56 | |
| *** jesusaur has quit IRC | 18:59 | |
| *** hasharDinner has joined #zuul | 18:59 | |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul-jobs master: Pull target from item in build image https://review.openstack.org/632528 | 19:00 |
| mordred | Shrews: ^^ that | 19:00 |
| Shrews | lol | 19:01 |
| mordred | tobiash: ^^ | 19:01 |
| *** ianw_pto is now known as ianw | 19:02 | |
| Shrews | mordred: hrm, no easy way to fix the opendevorg/python-builder upload without a change i guess? | 19:02 |
| mordred | yeah- I was just looking for something to change | 19:02 |
| tobiash | mordred: lgtm | 19:03 |
| mordred | tobiash: thanks! opendevorg/python-builder currently has the contents of opendevorg/python-base - which does not work very well as a builder image | 19:04 |
| tobiash | ah, so we need to land that and trigger a rebuild of it | 19:06 |
| Shrews | tobiash: https://review.openstack.org/632529 is the trigger, fwiw | 19:07 |
| Shrews | though perhaps it should use depends-on to prevent it from going in before the fix | 19:07 |
| tobiash | lol | 19:08 |
| pabelanger | Hmm, so I think swift container names have to be unique in cloud | 19:09 |
| pabelanger | confirming now | 19:10 |
| mordred | pabelanger: I think they only have to be unique within a project | 19:11 |
| pabelanger | mordred: not sure, i get Conflict (HTTP 409) (Request-ID: tx0000000000000000e4df3-005c476a63-3c95143d-default) if I try to create 'logs' container, but my project is empty | 19:13 |
| mordred | pabelanger: yay! I wonder if that's related to radosgw somehow | 19:13 |
| mordred | pabelanger: (I'm assuming this is on vexxhost) | 19:13 |
| pabelanger | mordred: yah, think we need help from mnaser here | 19:13 |
| timburke | mordred: that was my first thought, too :-) | 19:13 |
| pabelanger | yup | 19:13 |
| tobiash | pabelanger, mordred: they must be unique on the whole swift service | 19:14 |
| tobiash | it's the same on aws s3 btw | 19:14 |
| timburke | sure sounds like a ceph limitation to me ;-) | 19:14 |
| pabelanger | tobiash: yup, confirmed: https://object-storage-ca-ymq-1.vexxhost.net/v1/e0cc062046f0440db462b035c5df0910/vexxhost-poc-logs/2/2/d1a1139de90e4248fc42f1709da95685fc01287f/check/tox-linters/f193fa2/ | 19:14 |
| mordred | timburke: so you're saying that there is a bug in radosgw from a swift POV then? | 19:14 |
| pabelanger | so, i think our upload-logs-swift need to be updated to trap 409 resp, took me a while to figure this one out | 19:15 |
| mordred | tobiash: amazon s3 requires a container name to be unique across all of amazon? | 19:15 |
| pabelanger | as zuul would success but logs in swift be empty | 19:15 |
| tobiash | mordred: at least across the region | 19:15 |
| mordred | pabelanger: yah - seems like an important error to handle | 19:15 |
| mordred | tobiash: wow. that legitimately boggles my mind | 19:15 |
| timburke | mordred: yeah, the 409 description of "The container already exists under a different user’s ownership." from http://docs.ceph.com/docs/master/radosgw/swift/containerops/#http-response is *definitely* a divergence from swift behavior | 19:16 |
| tobiash | yes ;) | 19:16 |
| Shrews | wow, how wonderfully limiting | 19:16 |
| mordred | well - I'll give swift credit for doing the right thing here when s3 is clearly INSANE | 19:16 |
| * mordred hands timburke a cookie | 19:16 | |
| timburke | mordred: bonus fun: that sort of behavior has been implicated in some security holes, like https://www.safetydetective.com/blog/microsoft-outlook/ | 19:18 |
| mordred | timburke: uh - yeah. I'd imagine it would be - it's a **TERRIBLE** design | 19:19 |
| mordred | timburke: taht was literally the first attack that came to mind within the first few seconds of learning that container names have to be unique globally | 19:19 |
| pabelanger | best part, if you try to create 'logs' container in horizon, you actually get kicked out of horizon when you enter container name into web. something, something, security? | 19:20 |
| timburke | so protip: audit your CNAME records! make sure if any of them point to somewhere on AWS/Azure/wherever, they still actually exist and are under your control! | 19:21 |
| timburke | 'cause if you don't do it, someone else may ;-) | 19:21 |
| mordred | timburke: I'd say that if any of your CNAME records point to anything on AWS/Azure your'e already screwed | 19:21 |
| mordred | timburke: :) | 19:21 |
| pabelanger | so yay, I now have a functional zuul in vexxhost for 3pci testing, swift was the last step | 19:24 |
| mordred | pabelanger: woot! | 19:24 |
| pabelanger | and if people are intersted, it is even 1 zuul deploying another zuul via post pipeline: https://ansible-network.softwarefactory-project.io/zuul/builds?job_name=windmill-config-deploy | 19:25 |
| pabelanger | zuul all the way down | 19:25 |
| *** jesusaur has joined #zuul | 19:25 | |
| jkt | I'm getting a weird error, https://gerrit.cesnet.cz/c/ci/project-config/+/1277#message-ca7c4a1281c54ce48e4119844d6df0a9673f4d68 | 19:30 |
| jkt | I wonder why it says that it cannot find the "dummy" project | 19:30 |
| jkt | it is defined in https://gerrit.cesnet.cz/plugins/gitiles/ci/project-config/+/master/zuul/tenants.yaml#24 | 19:31 |
| pabelanger | jkt: have you reloaded zuul scheduler with latest main.yaml? | 19:32 |
| jkt | is it actually OK to use multiple tenants like this, with a shared config-projects repo? | 19:32 |
| tobiash | corvus: what do you expect if there are two pipelines with the same name in different config projects? | 19:32 |
| jkt | pabelanger: yes (it's via the script option, and I restarted the zuul-scheduler) | 19:32 |
| pabelanger | jkt: you should be able to look at debug log to see, possible scheduler didn't load the project for some reason. Eg: existing zuul.d folder | 19:33 |
| tobiash | I think I found a project that tried to 'redefine' a pipeline which is already defined in a central config repo | 19:33 |
| tobiash | and zuul doesn't seem to raise a config error | 19:33 |
| jkt | pabelanger: is it OK to reference the same project-config repo from multiple tenants like I am doing? | 19:33 |
| pabelanger | jkt: yes, but you won't be able to share things like secrets | 19:34 |
| pabelanger | so you need to be careful | 19:34 |
| jkt | my first tenant, the CzechLight-internal, indeed doesn't list the "dummy" repo | 19:34 |
| pabelanger | jkt: I think tobiash shares config-projects between tenants, currently I don't (plan to test) | 19:35 |
| jkt | tobiash: got a sample config I could look at? | 19:35 |
| tobiash | jkt: you can just add it normally to the tenant that should gate on the config repo, the other tenants can just list it by excluding 'project': http://paste.openstack.org/show/743115/ | 19:37 |
| jkt | tobiash: right, thanks, so this looks like trying to share two project definitions between two tenants is a no-go | 19:38 |
| tobiash | and yes, you can share secrets as the private keys are not dependent of the tenant | 19:38 |
| tobiash | jkt: not necessarily | 19:39 |
| tobiash | jkt: sharing check could be ok, but sharing gates is a no-go | 19:39 |
| jkt | I wonder if I actually *need* multi tenants in the first place | 19:39 |
| tobiash | jkt: depends on your role | 19:39 |
| jkt | I think that one benefit are separate dashboards *if* I can somehow block access to the top-level dashboard site | 19:40 |
| tobiash | if you're ci service provider for multiple distinct project it can make much sense | 19:40 |
| jkt | tobiash: my use case is that we have both public open source SW and some internal-still-proprietary SW within one Gerrit server | 19:40 |
| jkt | I would prefer not to share the CI logs etc with the rest of the world | 19:41 |
| tobiash | jkt: you can always put an authenticating proxy in front of zuul and even authenticate path based behind the tenant paths | 19:41 |
| *** hasharDinner is now known as hashar | 19:41 | |
| tobiash | jkt: ok, this use case needs much thoughts. If you can separate auth clearly by putting a proxy in front multi tenant would be an option | 19:42 |
| tobiash | but you have to also take log storage etc into account | 19:42 |
| jkt | ack | 19:42 |
| tobiash | so if you don't want to risk disclosure you also can think about two zuul deployments | 19:43 |
| tobiash | these still can share common projects if needed | 19:43 |
| openstackgerrit | Merged openstack-infra/zuul-jobs master: Pull target from item in build image https://review.openstack.org/632528 | 19:47 |
| corvus | tobiash: oh, hrm. i don't think you should be able to redefine a pipeline; i think that should be an error, and the first one should win. | 19:50 |
| *** pcaruana has joined #zuul | 19:50 | |
| tobiash | corvus: I just prototyped a quick test case and it seems that it doesn't throw an error and the last one wins | 19:51 |
| corvus | tobiash: i think we should fix that | 19:51 |
| tobiash | ++ | 19:51 |
| openstackgerrit | Merged openstack-infra/zuul-jobs master: Use become for packages in upload-puppetforge https://review.openstack.org/632163 | 20:08 |
| *** pcaruana has quit IRC | 20:11 | |
| *** pcaruana has joined #zuul | 20:24 | |
| mordred | Shrews: the python-builder updated image has been rebuilt and published - so if you docker pull it you should now be able to successfully build the nodepool images | 20:25 |
| Shrews | mordred: yeah, already doing so. looks good so far | 20:27 |
| pabelanger | corvus: did say you are looking to do a release of zuul this week? | 20:28 |
| pabelanger | I wouldn't mind one to pick up known_hosts / add_host fixes | 20:29 |
| jkt | tobiash, pabelanger: thanks, it was indeed due to my misunderstanding and reusing the same config in two tenants | 20:35 |
| jkt | with all the associated funny effects like duplicate job results, etc :) | 20:35 |
| openstackgerrit | Tobias Henkel proposed openstack-infra/zuul master: Raise an error if pipeline is defined twice https://review.openstack.org/632550 | 20:35 |
| corvus | yeah, sharing configuration is a spectrum -- there's no hard rule about whether you should share none, a little, or a lot between tenants. it depends on needs. | 20:39 |
| corvus | pabelanger: i think/hope we will have a release this week. | 20:39 |
| tobiash | there is also a node leak fix that should go into one of the next releases: https://review.openstack.org/613261 :) | 20:40 |
| corvus | tobiash: agreed; i will try to review it asap | 20:41 |
| tobiash | thanks :) | 20:41 |
| jkt | I wonder why a new job is not picked by Zuul here: https://gerrit.cesnet.cz/c/dummy/+/1286#message-8a7b11541f4dcfe1303705997a98f9bee5d5dcd9 | 20:58 |
| jkt | is that because I already have a noop job listed in my ci/project-config-public? | 20:59 |
| jkt | I do not have any base job defined (the docs said that there's an implicit one), and I have not defined any nodeset either | 21:01 |
| jkt | ...but defining a node set does not change stuff as far as I can tell | 21:03 |
| tobiash | jkt: the problem is that you don't have a job named base | 21:08 |
| tobiash | jkt: if you don't set a parent, the parent is by default 'base' | 21:08 |
| tobiash | jkt: and if that doesn't exist, it will be ignored | 21:08 |
| tobiash | jkt: but you can set 'parent: null' to make it a job without parent... | 21:09 |
| jkt | tobiash: thanks | 21:09 |
| corvus | tobiash, mordred, SpamapS, pabelanger: can you take a look at https://storyboard.openstack.org/#!/story/2004837 | 21:10 |
| jkt | tobiash: seems that https://softwarefactory-project.io/docs/zuul/glossary.html#term-base-job should be updated, though | 21:10 |
| tobiash | corvus: reading | 21:10 |
| corvus | jkt: there's some narrative documentation about base jobs in the quick start -- https://zuul-ci.org/docs/zuul/admin/quick-start.html#configure-zuul-pipelines then scroll down to "Every real job" (sorry, i guess we need better anchors in there) | 21:12 |
| corvus | jkt: https://zuul-ci.org/docs/zuul/admin/quick-start.html#configure-a-base-job is the real meat | 21:12 |
| jkt | corvus: I can do that, it's just that the docs speak about that "implicit base job" which is said to exist for each tenant | 21:13 |
| corvus | jkt: you may also be interested in the recommended base jobs https://git.zuul-ci.org/cgit/zuul-base-jobs/tree/ | 21:13 |
| jkt | if it exist, how come that I have to define it? | 21:13 |
| corvus | jkt: ah, i understand the ambiguity you are describing and agree we can/should update the docs to clarify it :) | 21:13 |
| openstackgerrit | Merged openstack-infra/nodepool master: Add dogpile.cache master to the -src tests https://review.openstack.org/625457 | 21:15 |
| jkt | corvus: I'll be happy to review it once it's done :), at this time I don't think I'm qualified to propose a better wording :) | 21:15 |
| corvus | the thing we need to convey: the site operator is expected to have defined at least one base job for the use of tenants and, if there is more than one, select one as the default for each tenant; once that's done, end-users can rely on that default being the implicit parent of any jobs which do not explicitly set a parent attribute. | 21:17 |
| jkt | just fyi, there's also no warning when there's no base job | 21:18 |
| corvus | that would be a very reasonable config error to report in the logs and web dashboard :) | 21:19 |
| *** pcaruana has quit IRC | 21:19 | |
| SpamapS | corvus: seems legit. There's a typo "emoved the allowed-projects acces restriction" | 21:25 |
| openstackgerrit | Tobias Henkel proposed openstack-infra/zuul master: Raise an error if pipeline is defined twice https://review.openstack.org/632550 | 21:37 |
| openstackgerrit | Merged openstack-infra/zuul master: Fix node leak when skipping child jobs https://review.openstack.org/613261 | 21:44 |
| corvus | okay, i'll push that change up now and make the story public | 21:58 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Set allowed-projects on untrusted jobs with secrets https://review.openstack.org/632566 | 21:59 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Set allowed-projects on untrusted jobs with secrets https://review.openstack.org/632566 | 22:01 |
| mordred | corvus: do you know if it's possible to consume 2 yaml anchors in the same place? | 22:07 |
| corvus | mordred: i don't know | 22:07 |
| SpamapS | corvus: aand I just remembered I can fix typos myself. ;) | 22:08 |
| mordred | corvus: I find that I want to do "vars: *zuul_image_vars + *nodepool_image_vars" - and I'm pretty sure that's not a thing | 22:08 |
| SpamapS | mordred: that's not a thing | 22:08 |
| corvus | when the allowed-projects change lands, i'll restart openstack with it, and if it checks out, i'll tag that as 3.5.0 | 22:08 |
| SpamapS | wanted that as well | 22:08 |
| SpamapS | and came up with spruce | 22:08 |
| mordred | corvus: ++ | 22:08 |
| SpamapS | as in, "oh you can't do that in YAML, you need something like spruce" | 22:09 |
| corvus | mordred: maybe i can get you on board the multiple inheritance train :) | 22:09 |
| mordred | corvus: I'm on the multiple inheritance train! I just don't understand it yet :) | 22:09 |
| SpamapS | mordred: you might be able to bend this one: https://yaml.org/type/merge.html | 22:09 |
| corvus | parent: [zuul-image-vars-job, nodepool-image-vars-job] | 22:10 |
| mordred | cooooh | 22:10 |
| mordred | corvus: (I realized that we need to build the zuul and the nodepool images using the dockerfile in the quick-start job) | 22:10 |
| corvus | mordred: yeah, i think i'll propose the change to do ^ that so that it's understandable :) | 22:10 |
| SpamapS | but merge YAML is like.. zomg wtf is this whenever I encounter it | 22:10 |
| mordred | corvus: and the best way to do that ... is with the build-docker-image role using the vars already defined | 22:10 |
| jkt | what can I use to SCP logs to some log server? The docs say that https://zuul-ci.org/docs/zuul-jobs/roles.html#role-upload-logs should do that, but doing something like zuul_logserver_root: "user@server:/path" results in an error | 22:11 |
| corvus | jkt, mordred: i think the upload-logs has an undocumented expectation that add_host will be used to add the remote server? | 22:12 |
| mordred | corvus: I think you might be right | 22:12 |
| jkt | corvus: I'm still embarasingly new to ansible; do you have an example? | 22:13 |
| corvus | jkt: so i think you need this role as well https://zuul-ci.org/docs/zuul-jobs/roles.html#role-add-fileserver | 22:13 |
| corvus | jkt: yes, here's openstack's use of that. lines 1-11 http://git.openstack.org/cgit/openstack-infra/project-config/tree/playbooks/base/post-logs.yaml | 22:14 |
| corvus | jkt: you can ignore the last task (lines 13+) | 22:14 |
| corvus | jkt: you may also want to omit line 5 for simplicity | 22:14 |
| mordred | SpamapS: ooh - I _could_ use that ... except I then need a list merge of a subkey :( | 22:15 |
| jkt | corvus: thanks, that makes sesne | 22:16 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul-jobs master: Document upload-logs reliance on add-fileserver https://review.openstack.org/632572 | 22:18 |
| corvus | jkt, mordred: ^ | 22:18 |
| corvus | er, hang on :) | 22:19 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul-jobs master: Document upload-logs reliance on add-fileserver https://review.openstack.org/632572 | 22:19 |
| corvus | mordred: that removes the {{site_logs}} stuff from the example. though i believe the intent was eventually to figure out how to get that to work with zuul-base-jobs. | 22:20 |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul master: Build zuul containers with dockerfile not pbrx https://review.openstack.org/631840 | 22:20 |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul master: Remove zuul-migrate and zuul-bwrap images https://review.openstack.org/632167 | 22:20 |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul master: Switch to zuul-jobs docker jobs https://review.openstack.org/632173 | 22:20 |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul master: Replace build-essential with gcc/g++ https://review.openstack.org/632576 | 22:20 |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul master: Update quick-start job to build images with dockerfile https://review.openstack.org/632577 | 22:20 |
| corvus | i'm not sure whether we should put that in the example now, or omit it to keep things simple. | 22:20 |
| mordred | well that was almost certainly not what I wanted to do | 22:20 |
| mordred | corvus: reading | 22:20 |
| jkt | corvus: I'll post my CR once I get this working (it's also about SSH keys and therefore I should first understand zuul's secrets, etc) | 22:22 |
| mordred | corvus: hrm. I'm also not sure ... | 22:24 |
| mordred | corvus: without the site_logs secret (or a secret with those things), the add_host command in the example is incomplete | 22:25 |
| mordred | corvus: since fileserver takes a dict | 22:25 |
| mordred | BUT - that's starting to get long ... but maybe that's a good place to put a more comprehensive chunk? | 22:25 |
| corvus | mordred: oh, then we should definitely -1 that and go with the full site_logs secret. :) | 22:27 |
| corvus | mordred: this is obvious now that you mention it. thanks, and sorry for the derp. | 22:27 |
| corvus | mordred: i'll rewrite that including site_logs | 22:27 |
| mordred | corvus: woot | 22:28 |
| SpamapS | mordred: I'd suggest wrapping your playbook in a zuul-friendly playbook and just passing the two lists in as separate vars. | 22:30 |
| SpamapS | I have a number of playbooks that work without zuul, and then a foo-zuul.yaml that just does an import_playbook: with the vars rearranged. | 22:30 |
| SpamapS | It's not a bad pattern really, keeps the zuul stuff out of the ansible. | 22:31 |
| mordred | SpamapS: yeah - I think I'm going to have to wind up doing something similarish | 22:34 |
| *** hashar has quit IRC | 22:35 | |
| corvus | yep, we could have an ansible playbook that "builds the zuul images" and then several jobs which invoke that, but it's also runnable locally | 22:38 |
| * mordred wishes he could do "docker build . --all-targets --tag-prefix=opendevorg" and have docker keep all of the intermediate images and tag them with {tag_prefix}/{intermediate-image-name} | 22:40 | |
| mordred | instead of needing call docker build 8 times - but what can you do | 22:41 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul-jobs master: Document upload-logs reliance on add-fileserver https://review.openstack.org/632572 | 22:41 |
| corvus | jkt, mordred: with more docs ^ | 22:41 |
| jkt | corvus: that looks good, thanks, I'll add my vote tomorrow | 22:44 |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul master: Update quick-start job to build images with dockerfile https://review.openstack.org/632577 | 22:55 |
| mordred | corvus: ^^ that is a much smaller version | 22:57 |
| *** ParsectiX has joined #zuul | 23:06 | |
| ParsectiX | Hi guys, | 23:11 |
| ParsectiX | I would like to introduce Zuul to my company. Although I don’t know from where to start. I have followed the quick start guide and played with the system. Documentation it’s really great but I’m not understanding all the moving parts yet. | 23:11 |
| ParsectiX | From where can I start ? Can you provide some guidance? | 23:11 |
| dmsimard | ParsectiX: ask away :D | 23:12 |
| dmsimard | there's plenty of zuul operators around | 23:12 |
| ParsectiX | Can I run/develop those jobs on my local machine? | 23:22 |
| pabelanger | corvus: just reading https://storyboard.openstack.org/#!/story/2004837 now, thanks | 23:24 |
| corvus | ParsectiX: one easy way would be to continue using the quick-start -- it's set up to run jobs on your local machine (in a container) | 23:25 |
| pabelanger | interesting issue, we aren't using 'restricted check' today, but have considered. | 23:26 |
| pabelanger | nice find | 23:26 |
| corvus | pabelanger: yeah, it's one of the original zuulv3 use cases, so if it doesn't work, it's a bug :| | 23:27 |
| pabelanger | yah | 23:27 |
| ParsectiX | Can I use Zuul without gerrit? | 23:29 |
| pabelanger | ParsectiX: yes, github is the other driver it supports | 23:31 |
| ParsectiX | Nice | 23:31 |
| corvus | ParsectiX: yes, you can use it with github. we hope to have other drivers in the future as folks volunteer to write them. | 23:31 |
| mordred | ++ | 23:31 |
| corvus | (you can also load configuration from plain git repos, but you can't use those for reviewing/gating changes) | 23:32 |
| openstackgerrit | Merged openstack-infra/zuul master: Set allowed-projects on untrusted jobs with secrets https://review.openstack.org/632566 | 23:36 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Fix secondary exception in fingergw https://review.openstack.org/632582 | 23:37 |
| corvus | pabelanger: ^ | 23:37 |
| corvus | pabelanger: http://logs.openstack.org/66/632566/2/gate/tox-py36/c927e1e/job-output.txt.gz#_2019-01-22_23_29_15_817739 | 23:37 |
| pabelanger | eep | 23:38 |
| pabelanger | +2 | 23:39 |
| ParsectiX | Can I use the same Ansible role to build a project locally and on the CI server? | 23:42 |
| corvus | ParsectiX: usually, yes. that's the idea behind zuul. there are some things that zuul doesn't allow untrusted projects to do (because they could compromise the zuul system itself). those usually only show up in fairly advanced playbooks. if you run into that, you can always have zuul run the 'ansible-playbook' command. | 23:44 |
| corvus | ParsectiX: but generally, the best approach is to write ansible playbooks that work on their own, and have zuul run them. basically, zuul supplies the inventory file. if you want to run them locally, you supply the inventory file. | 23:45 |
| ParsectiX | corvus: Thanks. I'm wondering how can I have the same output when running locally and on Zuul. Is there a way to integrate the callbacks and libraries to a local role? | 23:48 |
| SpamapS | You can run with the zuul plugins loaded, but there's no trivial way to do that... you have to set al the ansible variables just right to make it work. | 23:50 |
| SpamapS | I find myself wanting command output streaming all the time when running ansible locally. | 23:51 |
| corvus | ParsectiX: the streaming log output in particular won't work outside of zuul. we have plans to rework that and make it possible to integrate into upstream ansible, but that hasn't been fully implemented yet. | 23:51 |
| corvus | ParsectiX: but things like setting up log streaming and copying logs should be handled in base jobs, so they shouldn't require any zuul-specific tasks be added to the playbooks of normal jobs | 23:53 |
| ParsectiX | corvus: Yeah make sense. I just wanted to see if I can give to my developers the same user experience on both cases | 23:55 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!