| *** rlandy is now known as rlandy|bbl | 00:10 | |
| *** odyssey4me has quit IRC | 00:11 | |
| *** odyssey4me has joined #zuul | 00:11 | |
| openstackgerrit | Matthew Thode proposed openstack-infra/nodepool master: add gentoo systemd image for testing https://review.openstack.org/556661 | 00:17 |
|---|---|---|
| openstackgerrit | Matthew Thode proposed openstack-infra/nodepool master: DNM: testing gentoo nodepool-dsvm https://review.openstack.org/556679 | 00:18 |
| *** acozine1 has quit IRC | 00:21 | |
| *** flepied_ has quit IRC | 00:29 | |
| corvus | oh i still need to disable nullwrap. that would be good to do before release. | 00:38 |
| *** harlowja has quit IRC | 00:53 | |
| tristanC | corvus: thanks for answering the release question at the meeting | 01:03 |
| tristanC | what manual things you need to do to make the webui work? | 01:03 |
| *** tobiash has quit IRC | 01:05 | |
| corvus | tristanC: i'm not sure, morderd says he should have patches tomorrow, hopefully -- i'm fuzzy on the details | 01:24 |
| tristanC | fwiw, having https://review.openstack.org/551327 and the patch bellow would make webui developper life easier | 01:36 |
| openstackgerrit | Matthew Thode proposed openstack-infra/nodepool master: add gentoo systemd image for testing https://review.openstack.org/556661 | 01:58 |
| openstackgerrit | Matthew Thode proposed openstack-infra/nodepool master: DNM: testing gentoo nodepool-dsvm https://review.openstack.org/556679 | 01:59 |
| *** Wei_Liu has joined #zuul | 02:13 | |
| *** dmsimard|cave is now known as dmsimard | 02:29 | |
| *** rlandy|bbl is now known as rlandy | 03:19 | |
| *** rlandy has quit IRC | 03:19 | |
| *** tobiash has joined #zuul | 03:49 | |
| *** bhavik1 has joined #zuul | 03:59 | |
| openstackgerrit | Matthew Thode proposed openstack-infra/nodepool master: add gentoo systemd image for testing https://review.openstack.org/556661 | 04:23 |
| openstackgerrit | Matthew Thode proposed openstack-infra/nodepool master: DNM: testing gentoo nodepool-dsvm https://review.openstack.org/556679 | 04:24 |
| *** bhavik1 has quit IRC | 04:57 | |
| *** Wei_Liu has quit IRC | 05:35 | |
| *** Wei_Liu has joined #zuul | 05:36 | |
| tobiash | corvus: timezone magic happened :) | 06:38 |
| tobiash | corvus, jlk: I retested the latest github3.py in my testenv and startup, check and gating works without exceptions in the log | 06:38 |
| *** chrnils has joined #zuul | 06:44 | |
| *** flepied has joined #zuul | 07:24 | |
| *** xinliang has joined #zuul | 07:51 | |
| *** xinliang has quit IRC | 07:51 | |
| *** xinliang has joined #zuul | 07:51 | |
| *** tosky has joined #zuul | 07:54 | |
| LinuxJedi | Shrews (or anyone else): do you happen to have an example job that will execute on the static nodepool? All the examples appears to be OpenStack based and I appear to be missing some key points in my yaml | 08:34 |
| tobiash | LinuxJedi: what do you have right now? using static nodes should not be different than any other node | 08:47 |
| LinuxJedi | https://github.com/libattachsql/libattachsql/pull/202/files and https://github.com/libattachsql/libattachsql-zuul not sure what I'm doing wrong as I've never done this before but "scanbuild" isn't getting triggered | 08:48 |
| LinuxJedi | my nodepool is based off Shrews' one if that helps: http://paste.openstack.org/show/714027/ | 08:50 |
| tobiash | LinuxJedi: scanbuild uses the ubuntu-xenial label which might be a different label than the label you defined in nodepool for your static node | 08:50 |
| tobiash | https://github.com/libattachsql/libattachsql-zuul/blob/master/zuul.yaml#L8 | 08:50 |
| tobiash | ah ok, so the label matches | 08:50 |
| tristanC | LinuxJedi: is this project a "config-projects" or "untrusted-projects" ? | 08:51 |
| tobiash | LinuxJedi: do you have libattachsql/libattachsql-zuul listed on your tenant config? | 08:52 |
| tristanC | LinuxJedi: if it's a config-project, then the zuul.yaml is only taken into account when the change is merged | 08:52 |
| LinuxJedi | my main.yaml: http://paste.openstack.org/show/715360/ | 08:53 |
| tobiash | hrm, I don't see any obvious mistake there | 08:53 |
| tobiash | LinuxJedi: can you do a recheck and look at the logs of the zuul-scheduler? | 08:53 |
| LinuxJedi | tristanC: I tried shoving it all in libattachsql's zuul.yaml as an untrusted-project and it complained that I couldn't define jobs that (I think, that was yesterday) | 08:54 |
| openstackgerrit | Ian Wienand proposed openstack-infra/zuul-jobs master: Handle blank public_ipv4 https://review.openstack.org/556784 | 08:54 |
| LinuxJedi | scheduler log: http://paste.openstack.org/show/715362/ | 08:55 |
| tobiash | LinuxJedi: can you change the log level to debug and repeat that? | 08:57 |
| tobiash | the info logging doesn't seem to be helpful in this case | 08:57 |
| LinuxJedi | tobiash: how do I do that? | 08:57 |
| tobiash | do you have a log config file? | 08:57 |
| LinuxJedi | what is one of those? | 08:57 |
| LinuxJedi | (think I have had about 6 hours experience of Zuul and only understood the quickstart :) ) | 08:58 |
| tobiash | LinuxJedi: https://docs.openstack.org/infra/zuul/admin/components.html#attr-scheduler.log_config | 08:58 |
| tobiash | you need to configure the logging config | 08:58 |
| tobiash | let me quick get a sample logging config | 08:59 |
| tobiash | LinuxJedi: that's the logging config I am using: http://paste.openstack.org/show/715364/ | 09:01 |
| tobiash | it's a bit messy currently but should get you going | 09:02 |
| LinuxJedi | tobiash: awesome! Thanks, how about this? http://paste.openstack.org/show/715366/ | 09:04 |
| tobiash | that's at least more logging :) | 09:05 |
| tobiash | LinuxJedi: do you have a base job? | 09:06 |
| tobiash | 2018-03-27 10:03:39,390 DEBUG zuul.layout: No matching parents for job base and change <Change 0x7f44341d0470 202,809001883c3a356346363051ad84e087cdfa44f0> | 09:06 |
| LinuxJedi | I don't think so | 09:07 |
| tobiash | then that's the problem | 09:07 |
| * LinuxJedi quickly googles what that is :) | 09:08 | |
| tobiash | LinuxJedi: you probably missed a step in the guide almost at the end | 09:09 |
| tobiash | search for 'Edit /etc/zuul/main.yaml so that it looks like this:' in the zuul from scratch guide | 09:09 |
| tobiash | LinuxJedi: you need to add openstack-infra/zuul-base-jobs and openstack-infra/zuul-jobs to your main.yaml | 09:10 |
| LinuxJedi | ah! Ok. I missed that out as it looked OpenStack specific. My bad! | 09:12 |
| *** hashar has joined #zuul | 09:17 | |
| LinuxJedi | progress! Scanbuild now shows as queued at least :) | 09:20 |
| tobiash | yah, that's like a standard library for zuul | 09:20 |
| LinuxJedi | many thanks tobiash :) | 09:20 |
| tobiash | no problem :) | 09:20 |
| LinuxJedi | "nodepool.NodePool: ZooKeeper suspended. Waiting" <- that would explain my next problem :) | 09:24 |
| tobiash | LinuxJedi: is zookeeper running and nodepool configured to use that zookeeper? | 09:27 |
| LinuxJedi | yes and yes (localhost). Zookeeper's logging appears to be broken so not sure what is going on there | 09:28 |
| LinuxJedi | ok, fixed that. Turns out I missed the port on nodepool's zookeeper configuration. We have progress! A failure! Appears I have some kind of key problem with the static builder. Time to read up on how that is supposed to work :) | 09:51 |
| *** tosky__ has joined #zuul | 10:25 | |
| *** tosky is now known as Guest72548 | 10:25 | |
| *** tosky__ is now known as tosky | 10:25 | |
| LinuxJedi | tobiash: sorry to bug you again, any idea what the "implicit role" is? http://paste.openstack.org/show/715378/ | 10:49 |
| tobiash | LinuxJedi: do you have an issue or is it just the log lines? | 10:50 |
| tobiash | they should be harmless | 10:50 |
| tobiash | zuul adds an implicit roles path to ansible if your repo under test contains a /roles dir | 10:51 |
| tobiash | if it doesn't it spits out this log line | 10:51 |
| *** JasonCL has quit IRC | 10:51 | |
| tobiash | maybe we want to change that line to debug log level | 10:51 |
| tobiash | corvus: am I correct? ^ | 10:51 |
| LinuxJedi | tobiash: ok. I thought that might be my problem. Basically executor doesn't appear to be executing, retries and fails. The worker is showing "Mar 27 11:51:27 ubuntu sshd[3264]: Connection closed by 192.168.56.101 port 41308 [preauth]" | 10:52 |
| tobiash | LinuxJedi: execute 'zuul-executor verbose' | 10:52 |
| tobiash | that should make the zuul-executor ansible log verbose | 10:52 |
| tobiash | maybe that helps in debugging | 10:52 |
| tobiash | but your error message might indicate that you have a permission problem | 10:53 |
| tobiash | does your static node have a zuul user with public key auth using the zuul ssh key? | 10:53 |
| LinuxJedi | I suspect the ssh keys are where I went wrong. I couldn't find any good documentation on which key should go where. I suspect I need the nodepool public key as an authorized key on the static node? | 10:55 |
| tobiash | yes | 10:55 |
| LinuxJedi | better, still failing but SSH auth is at least happening now :) | 10:57 |
| *** sshnaidm has quit IRC | 11:30 | |
| LinuxJedi | Hmmm... the ansible verbose logging isn’t giving me anything so I guess it is something else | 11:31 |
| *** openstackgerrit has quit IRC | 11:33 | |
| Shrews | morning | 12:04 |
| Shrews | LinuxJedi: i can be of assistance after some coffee | 12:04 |
| Shrews | LinuxJedi: did you set the host keys? | 12:05 |
| Shrews | oh, you're past that point. /me coffees | 12:07 |
| *** sshnaidm has joined #zuul | 12:09 | |
| LinuxJedi | lol! Yep. Past that. At the point where execution is failing but not really telling me why | 12:27 |
| *** openstackgerrit has joined #zuul | 12:27 | |
| openstackgerrit | Jens Harbott (frickler) proposed openstack-infra/zuul-jobs master: Handle blank public_ipv4 https://review.openstack.org/556784 | 12:27 |
| Shrews | LinuxJedi: i had to manually install python2 on my ubuntu VMs... but can i see your latest zuul debug log? | 12:28 |
| LinuxJedi | let me re-run and I'll give you what I have | 12:28 |
| Shrews | LinuxJedi: and seeing your zuul.yaml in your trusted and untrusted projects would help | 12:28 |
| *** Wei_Liu has quit IRC | 12:30 | |
| *** Wei_Liu has joined #zuul | 12:30 | |
| LinuxJedi | Shrews: debug log and execution log: http://paste.openstack.org/show/715386/ | 12:30 |
| LinuxJedi | Shrews: untrusted: https://github.com/libattachsql/libattachsql/pull/202/files config: https://github.com/libattachsql/libattachsql-zuul | 12:30 |
| Shrews | oh, you've defined your job in the config project. you could do that in the untrusted | 12:33 |
| LinuxJedi | does it make a difference? | 12:34 |
| LinuxJedi | I'll move it and see... | 12:35 |
| *** rlandy has joined #zuul | 12:36 | |
| Shrews | LinuxJedi: probably not. also doesn't look like the executor is using DEBUG | 12:38 |
| Shrews | as there is frustratingly little info there | 12:39 |
| LinuxJedi | ooo... Adding Python to the worker appears to have fixed things a bit... | 12:40 |
| Shrews | LinuxJedi: we define jobs in our trusted config. but the idea is that each untrusted repo can define their own | 12:41 |
| LinuxJedi | Shrews: ok, so the build failed. I'm pretty sure I know why it failed, but I'm not getting any build logs show in builds.html | 12:42 |
| Shrews | LinuxJedi: did you get debug enabled on the executor? | 12:42 |
| LinuxJedi | not yet, still figuring out the magic for that | 12:43 |
| *** odyssey4me has quit IRC | 12:43 | |
| Shrews | LinuxJedi: i use this locally: http://paste.openstack.org/show/715391/ | 12:43 |
| *** odyssey4me has joined #zuul | 12:43 | |
| Shrews | everything goes to /tmp/zuul-debug.log | 12:44 |
| Shrews | LinuxJedi: can use that for both scheduler and executor | 12:44 |
| LinuxJedi | ah! Stupid me, I was looking for a way to do it in the logging config file rather than zuul.conf... | 12:45 |
| Shrews | LinuxJedi: zuul.conf should have something like: http://paste.openstack.org/show/715393/ | 12:46 |
| LinuxJedi | 👍 yep, it is now. Will retest and get some debugging info... | 12:48 |
| Shrews | oh neat. weechat rendered that thumbs up | 12:49 |
| Shrews | how 2000's of it | 12:49 |
| LinuxJedi | lol! I'm impressed that IRCCloud took colon-plus-colon as thumbs up like Slack does | 12:50 |
| LinuxJedi | colon-plus-one-colon even | 12:51 |
| LinuxJedi | Shrews: latest debug log, better? http://paste.openstack.org/show/715394/ :) | 12:51 |
| Shrews | LinuxJedi: well, it's cutoff, but looks like ansible was doing stuff | 12:52 |
| LinuxJedi | Shrews: yep. Installing Python on the worker fixed that | 12:53 |
| Shrews | \o/ | 12:53 |
| *** dkranz has joined #zuul | 13:02 | |
| dmsimard | Is it expected that Zuul doesn't enqueue a child patch when a parent patch (depends-on) has merged ? | 13:07 |
| AJaeger | dmsimard: yes - if those do not co-gate. If tehy share the same gate queue, then they are submitted together | 13:08 |
| *** JasonCL has joined #zuul | 13:11 | |
| *** JasonCL has quit IRC | 13:12 | |
| *** JasonCL has joined #zuul | 13:13 | |
| tobiash | dmsimard: you can add parent-change-enqueued as trigger | 13:14 |
| tobiash | dmsimard: https://docs.openstack.org/infra/zuul/admin/drivers/zuul.html#trigger-configuration | 13:14 |
| *** JasonCL has quit IRC | 13:15 | |
| *** JasonCL has joined #zuul | 13:15 | |
| LinuxJedi | so, I'm up to the point where things are building. But the builds tab in the Zuul dashboard is empty. Is there some kind of config to turn that on? | 13:16 |
| Shrews | tobiash: does that ^^ require an sql database? | 13:21 |
| tobiash | Shrews, LinuxJedi: yes | 13:21 |
| Shrews | I absolutely *love* how that isn't documented anywhere, btw | 13:21 |
| tobiash | you need to configure an sql connection and add sql reporting to the pipeline | 13:21 |
| LinuxJedi | do I need to install any schemas or is that automatic? | 13:21 |
| tobiash | that is automatic | 13:22 |
| LinuxJedi | oh, in the pipeline as well as the config... OK, I'll try and find that too | 13:22 |
| tobiash | LinuxJedi: the pipeline part is there: http://git.openstack.org/cgit/openstack-infra/project-config/tree/zuul.d/pipelines.yaml#n49 | 13:23 |
| tobiash | and the config part: https://docs.openstack.org/infra/zuul/admin/drivers/sql.html | 13:24 |
| *** acozine1 has joined #zuul | 13:29 | |
| * mordred waves to LinuxJedi | 13:30 | |
| LinuxJedi | hey mordred :) | 13:30 |
| Shrews | Hey, who wants to create a database designed for use in the cloud???? | 13:35 |
| Shrews | If we just had an energetic Aussie and a scrappy QA person, I think we could do it. | 13:38 |
| LinuxJedi | lol! | 13:38 |
| LinuxJedi | ok, well, I got the builds tab working, but the logs give me URLs like "finger://localhost.localdomain/e4e737439a584d699822a05a1174ef4a" instead of.. logs | 13:39 |
| tobiash | mordred: do you know if shade's create_image_snapshop will abort with wait=true and the in progress image disappears? | 13:41 |
| *** dtruong2_ has joined #zuul | 13:42 | |
| *** elyezer has quit IRC | 13:49 | |
| corvus | tobiash: yes, let's switch the implied role line to debug :) | 13:50 |
| LinuxJedi | I'm afraid I think I'm not going to want to know the answer to this, but how do you make the log_url be an actual web URL I can retrieve the logs from? | 13:51 |
| mordred | tobiash: hrm. that's a fascinating question | 13:52 |
| mordred | tobiash: nope. it will not abort | 13:53 |
| corvus | LinuxJedi: i think the first question is: where do you want the logs to go? | 13:53 |
| corvus | LinuxJedi: (also, you've now run past the end of the zuul-from-scratch doc -- you're pretty close to a working system, but we're definitely in undocumented territory) | 13:54 |
| *** dtruong2_ has quit IRC | 13:54 | |
| mordred | tobiash: it will happily continue polling for the image to complete - and will assume any errors are temporary | 13:54 |
| corvus | LinuxJedi: we copy our logs to a static webserver which serves them via apache. | 13:54 |
| corvus | LinuxJedi: for testing, it can even be the same server zuul runs on, but obviously, that's probably not what you want in the long run | 13:54 |
| mordred | tobiash: this is partly because we don't want an API service hiccup to cause us to think the image failed | 13:54 |
| corvus | LinuxJedi: (in the future, we're going to move our log hosting to an object store, like swift) | 13:55 |
| tobiash | mordred: interesting | 13:55 |
| mordred | tobiash: that said - if we got an image id in the first place, we could treat 404 from the wait_for_image loop as an error - like a 404 shouldn't happen in the middle of waiting | 13:55 |
| LinuxJedi | corvus: ok, that is fine, I can work with that. Where are they stored? | 13:55 |
| corvus | LinuxJedi: did you add the zuul-base-jobs repo? | 13:56 |
| LinuxJedi | yes | 13:57 |
| corvus | LinuxJedi: did you also add the zuul-jobs repo? | 13:58 |
| LinuxJedi | yes | 13:59 |
| LinuxJedi | I needed to do that otherwise builds didn't work at all | 13:59 |
| tristanC | tobiash: do you have a release pipeline working with github? | 14:00 |
| tobiash | tristanC: a pipeline triggered by a tag? | 14:00 |
| pabelanger | morning | 14:01 |
| pabelanger | https://zuul-ci.org/docs/zuul/user/jobs.html seems to 404, is that a known issue? | 14:01 |
| *** elyezer has joined #zuul | 14:01 | |
| tristanC | tobiash: yes, not sure how to filter on ^refs/tags/ for github push | 14:01 |
| *** jamielennox has quit IRC | 14:01 | |
| *** jamielennox has joined #zuul | 14:02 | |
| tobiash | tristanC: I have this: http://paste.openstack.org/show/715406/ | 14:02 |
| tobiash | but I'm not sure somebody tried it out already ;) | 14:02 |
| mordred | tobiash: remote: https://review.openstack.org/556865 Abort waiting for image on 404 | 14:03 |
| mordred | tobiash: something like that (untested) | 14:03 |
| corvus | mordred: were we going to do any default log publishing in zuul-base-jobs? | 14:03 |
| mordred | corvus: we'd talked about it | 14:04 |
| mordred | corvus: I think witha default of publishing to a local directory | 14:04 |
| corvus | mordred: yeah, i think that's probably the right thing to do (as i sit here and type out all the stuff LinuxJedi will need to do) | 14:04 |
| mordred | corvus: so that an AIO install using zuul-base-jobs would wind up with logs in a known directory (that could even be served by a default aio apache config) | 14:04 |
| mordred | corvus: ++ | 14:05 |
| tristanC | tobiash: i see, thanks, i'll give it a try | 14:06 |
| tobiash | tristanC: but I have no idea if it works ;) | 14:07 |
| corvus | LinuxJedi: we'll make the zuul-base-jobs repo more useful in a bit. but for now, to keep moving, you'll need to make your own base job. start by removing zuul-base-jobs from your tenant config, then do the steps in this etherpad: https://etherpad.openstack.org/p/sqJ1DzLmSU | 14:07 |
| corvus | LinuxJedi: (replace "config-repo" with whatever you called your config-repo) | 14:08 |
| corvus | LinuxJedi: that copies the base job, and its playbooks, from zuul-base-jobs into your config-repo. | 14:08 |
| corvus | LinuxJedi: then, once you have your own base job, we can add log publishing to it | 14:08 |
| corvus | oh, actually, the upload-logs role uses ssh doesn't it? | 14:09 |
| corvus | oh, it uses synchronize -- maybe that will work if we just run it on localhost... | 14:11 |
| corvus | mordred: what do you think of the etherpad? ^ | 14:11 |
| openstackgerrit | Tobias Henkel proposed openstack-infra/zuul master: Switch implicit role not found message to debug https://review.openstack.org/556866 | 14:12 |
| tobiash | corvus, LinuxJedi ^ | 14:12 |
| mordred | corvus: I think it looks fine - assuming synchronize will dtrt when run on localhost | 14:13 |
| corvus | LinuxJedi: okay, let me know how that goes :) | 14:15 |
| pabelanger | I'm trying to help ansible-network bring some jobs online with softwarefactory, but unsure why zuul is reporting a configuration error to an across projects. | 14:23 |
| pabelanger | eg: https://github.com/ansible-network/ansible-zuul-jobs/pull/3#issuecomment-376540763 | 14:23 |
| corvus | LinuxJedi, Shrews: i'm thinking i'd like to add run-time and config-time options to switch the logging from standard to debug. so basically instead of having one built-in logging config, we have two. | 14:23 |
| pabelanger | would have expected that to be reported to ansible-network/network-engine | 14:24 |
| corvus | pabelanger: oy, that formats very poorly :( | 14:24 |
| tobiash | corvus: yah, that makes sense if you don't want to have some fully fledged logging config file | 14:27 |
| corvus | pabelanger: take it at face value | 14:27 |
| corvus | pabelanger: network-engine really has that config: https://github.com/ansible-network/network-engine/blob/ansible-test/.zuul.yaml | 14:28 |
| pabelanger | Oh, somebody force merged code | 14:28 |
| pabelanger | corvus: thanks, I missed that | 14:28 |
| corvus | pabelanger: of course, why the proposed change to ansible-zuul-jobs didn't fix it, i'm not sure | 14:28 |
| corvus | pabelanger: can i see the zuul tenant config? | 14:30 |
| pabelanger | corvus: let me find it | 14:30 |
| pabelanger | but, that branch likely shouldn't exist either | 14:30 |
| tristanC | corvus: the tenant configuration is https://softwarefactory-project.io/r/gitweb?p=config.git;a=blob;f=zuul/ansible_networking.yaml | 14:31 |
| pabelanger | thanks | 14:31 |
| corvus | tristanC, pabelanger: i'm very confused. afaict, zuul should currently have a broken config because of the missing job, but that PR should have fixed it... | 14:38 |
| pabelanger | corvus: I think the PR is the broken code, if I am understanding write | 14:38 |
| pabelanger | right* | 14:39 |
| corvus | pabelanger: what's broken about it? | 14:39 |
| pabelanger | https://github.com/ansible-network/ansible-zuul-jobs/tree/ansible-test-sanity was created instead of using personal fork | 14:39 |
| pabelanger | then the PR to master was created from it | 14:40 |
| corvus | i'm not sure how that causes the job not to be defined (that should cause it to be defined on two branches) | 14:41 |
| pabelanger | ah, I think I understand what you are saying | 14:42 |
| corvus | pabelanger, tristanC: i'd probably look at the debug log at this point to try to figure out what i'm missing about what the configloader was doing | 14:43 |
| pabelanger | corvus: ack, let me see about getting them | 14:43 |
| tristanC | corvus: sorry about the confusion, it's because we are running with: https://review.openstack.org/#/c/535511/ | 14:45 |
| tristanC | or i think it is... the project errors are not kept in the tenant, and it can be observed on https://ansible.softwarefactory-project.io/zuul/info loading_errors field | 14:46 |
| LinuxJedi | corvus: ok, doing that etherpad change the URLs in the web front end, but it doesn't appear to have put logs in /srv/static/logs | 14:48 |
| corvus | tristanC: that looks like the same error -- my question still remains -- why doesn't the PR fix the error? (it adds the job) | 14:50 |
| jlk | tobiash: that's good news re github3.py | 14:52 |
| corvus | LinuxJedi: okay, if you run "zuul-executor keep", it will tell the executor not to delete the build directory after the job is complete. so if you do that, and then run the job again, you can look in tho log to see where the build directory is, then cd into that and look at the logs. should be in .../work/logs/job-output.txt and job-output.json. | 14:53 |
| corvus | LinuxJedi: that should help us figure out why the log copy didn't work | 14:54 |
| LinuxJedi | 👍 | 14:54 |
| LinuxJedi | @corvus helpful... 2018-03-27 14:55:52.711059 | TASK [upload-logs : Upload logs to log server 2018-03-27 14:55:53.102382 | localhost | Output suppressed because no_log was given | 14:58 |
| corvus | ha | 14:58 |
| corvus | mordred: i'm assuming no_log is set on the synchronize task in upload-logs just because it's chatty? there's no security implication? | 15:00 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul-jobs master: Add a flag to disable no_log on upload-logs https://review.openstack.org/556877 | 15:06 |
| corvus | mordred, LinuxJedi: ^ | 15:06 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul-jobs master: Add a flag to disable no_log on upload-logs https://review.openstack.org/556877 | 15:12 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul-jobs master: DNM: Test change to upload-logs https://review.openstack.org/556885 | 15:13 |
| *** JasonCL has quit IRC | 15:23 | |
| tobiash | corvus: there should be no security implication | 15:24 |
| *** JasonCL has joined #zuul | 15:24 | |
| tobiash | I wonder if there is a way to make it non verbose without setting no log | 15:25 |
| corvus | tobiash: good question. i think the issue is that synchronize sets the rsync flag which outputs info about every file. maybe it's possible to just disable that flag | 15:26 |
| *** JasonCL has quit IRC | 15:27 | |
| tobiash | corvus: a quick scan of the docs seem to tell me there is none unless there is a way to make disable verbose in rsync with a custom flag | 15:27 |
| corvus | i think it sets the --out-format flag, and does so in order to determine whether the task should set changed or not | 15:31 |
| *** JasonCL has joined #zuul | 15:33 | |
| corvus | i'm not seeing an obvious way to avoid it | 15:34 |
| tobiash | Hm, changing that to command and invoking rsync manually would probably solve that but that's a bit ugly | 15:36 |
| corvus | heh, yeah | 15:36 |
| *** JasonCL has quit IRC | 15:36 | |
| *** JasonCL has joined #zuul | 15:36 | |
| tobiash | But well, even the ansible docs state: 'synchronize is not intended to provide access to the full power of rsync, but does make the most common invocations easier to implement. You stillmay need to call rsync directly via command or shell depending on your use case.' | 15:38 |
| tobiash | If we have that 'unverbose but please tell me the errors' use case it might be ok to use rsync directly | 15:39 |
| *** JasonCL has quit IRC | 15:43 | |
| *** JasonCL has joined #zuul | 15:43 | |
| *** JasonCL has quit IRC | 15:48 | |
| Shrews | corvus: how about we update the topic to point things to zuul-ci.org? | 15:53 |
| corvus | Shrews: we still need to trigger docs publishing | 15:53 |
| Shrews | corvus: oh, i saw the docs already there and thought it was done | 15:53 |
| corvus | maybe i'll manually enqueue some refs later to get them all published | 15:55 |
| pabelanger | corvus: tristanC: to loop back to before, it was because we had zuul.d/job.yml for zuul configuration (not the file extension). Changing to .yaml worked as expected | 15:55 |
| pabelanger | thanks to tristanC for noticing it | 15:55 |
| pabelanger | note* | 15:55 |
| corvus | pabelanger: ah, that's a behavior difference introduced with the load-broken-config patch. a stock zuul would not have left an error comment on that change. | 15:56 |
| corvus | (because it would not have been detected as a config change) | 15:56 |
| pabelanger | corvus: yes, I agree | 15:57 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul-jobs master: Add a flag to disable no_log on upload-logs https://review.openstack.org/556877 | 16:17 |
| openstackgerrit | James E. Blair proposed openstack-infra/zone-zuul-ci.org master: Add git.zuul-ci.org https://review.openstack.org/556964 | 16:23 |
| *** JasonCL has joined #zuul | 16:23 | |
| corvus | Shrews: i'm triggering builds for the missing docs now | 16:25 |
| corvus | Shrews: there's a significant backlog, so that may not happen for a while :( | 16:29 |
| Shrews | *nod* | 16:30 |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul master: Reorganize REST API and dashboard urls https://review.openstack.org/556967 | 16:31 |
| Shrews | corvus: I'm trying to find the best way to present the gerrit version of the quick start guide. I've got notes for ubuntu, but the other guide does fedora. I also didn't bother with "Service Management" stuff for my test. | 16:34 |
| Shrews | corvus: So the guides will be inconsistent in at least those aspects | 16:35 |
| Shrews | I just ran things (zookeeper, gerrit, nodepool, zuul, geard) by hand, but maybe I should present a more "typical" setup? | 16:36 |
| corvus | Shrews: that should all be the same | 16:36 |
| Shrews | Or present it as a "here is a guide to experiment locally" | 16:36 |
| pabelanger | corvus: I wasn't sure if you see my question from this morning, but https://zuul-ci.org/docs/zuul/user/jobs.html is 404. Wanted to see if that was a known issue | 16:36 |
| pabelanger | https://zuul-ci.org/docs/zuul/ seems okay | 16:36 |
| corvus | Shrews: i'd suggest we expand the zuul-from-scratch guide to branch. so it starts out with setting everything up in general, then, we have a section for setting up gerrit stuff, and an alternative section for setting up github stuff | 16:37 |
| corvus | pabelanger: we're waiting on publishing jobs to run, will probably be (much) later today | 16:38 |
| pabelanger | ack | 16:38 |
| mordred | corvus, pabelanger, Shrews, tobiash: remote: https://review.openstack.org/556973 Support updated REST and HTML layout <-- that is a companion to //review.openstack.org/556967 | 16:44 |
| mordred | this is the breaking change that was mentioned in the meeting yesterday | 16:44 |
| corvus | tristanC: ^ | 16:44 |
| openstackgerrit | Paul Belanger proposed openstack-infra/nodepool master: WIP: Add debian-sid to nodepool-dsvm testing https://review.openstack.org/556974 | 16:45 |
| openstackgerrit | Paul Belanger proposed openstack-infra/nodepool master: DNM: testing debian-sid https://review.openstack.org/556975 | 16:45 |
| openstackgerrit | Paul Belanger proposed openstack-infra/nodepool master: DNM: testing debian-sid https://review.openstack.org/556975 | 16:49 |
| corvus | mordred: did you run tests on https://review.openstack.org/556967 ? i'm wondering whether the web url tests we wrote are going to pass without changes. and if so... maybe we need to test some more urls? | 16:50 |
| mordred | tristanC, corvus, tobiash, Shrews, pabelanger: the zuul patch is incomplete - needs some updates to the js, I think to the tests - and I want to add in some deploymnet docs - but I wanted to get the changes up for people's eyeballs early | 16:50 |
| mordred | corvus: I did not - and I'll be surprised if they do work :) | 16:50 |
| corvus | mordred: aha! sounds good | 16:50 |
| *** hashar is now known as hasharDockerized | 16:52 | |
| *** JasonCL_ has joined #zuul | 17:08 | |
| *** JasonCL has quit IRC | 17:12 | |
| openstackgerrit | Merged openstack-infra/zone-zuul-ci.org master: Add git.zuul-ci.org https://review.openstack.org/556964 | 17:15 |
| openstackgerrit | David Shrewsbury proposed openstack-infra/zuul master: WIP: Reorganzie ZFS document https://review.openstack.org/556988 | 17:26 |
| Shrews | corvus: Not quite complete, but sort of what I'm thinking ^^^ based on your suggestion | 17:26 |
| Shrews | would similarly split up nodepool config based on driver (openstack vs. static) | 17:28 |
| corvus | Shrews: cool, broadly speaking i think that'll be good; i'll take a closer look after the build job completes | 17:29 |
| Shrews | yeah, hard to see without viewing in a browser | 17:29 |
| Shrews | corvus: mainly just making sure this is the direction you'd like to see before changing too much | 17:30 |
| corvus | Shrews: yep; i suspect my comments will be minor, if any. | 17:31 |
| corvus | pabelanger, mordred: i believe https://review.openstack.org/556877 is ready to merge. the change that tests it sucessfully uploaded logs: https://review.openstack.org/556885 | 17:34 |
| clarkb | corvus: re ^ and also the ipv6 thing happening in -infra, it would be really neat if ansible had something like set -x | 17:35 |
| corvus | LinuxJedi: when https://review.openstack.org/556877 merges, you can restart your zuul scheduler, then add the "zuul_log_verbose" line in the etherpad https://etherpad.openstack.org/p/sqJ1DzLmSU | 17:36 |
| corvus | clarkb: i usually think ara serves that purpose | 17:36 |
| clarkb | corvus: it sort of does, in the ipv6 case it doesn't actually show you what the value that it evaluated is, just that it resulted in false (wheras in bash conditionals are shown with set -x) | 17:37 |
| clarkb | I guess verbosity in synchronize is less about control flow | 17:37 |
| corvus | yeah, that's a choice we made | 17:37 |
| corvus | we just need to make the choice reversible :) | 17:38 |
| LinuxJedi | corvus: Cool, thanks. May be tomorrow when I can get back to it | 17:38 |
| corvus | clarkb, dmsimard: i wounder if there's a way to access the values of variables used in conditions inside a callback plugin. then maybe ara/json can record them | 17:38 |
| dmsimard | corvus: it's tricky because variables aren't static, you can likely get the value of a variable at one point in time but there's no guarantee that it will remain the same throughout the playbook (or even on different hosts) | 17:39 |
| corvus | LinuxJedi: ack; sorry this took a bit -- our testing process for changing a global role like that has a few steps so we don't regress. at least it should be in place for you tomorrow | 17:39 |
| pabelanger | corvus: looking | 17:39 |
| corvus | dmsimard: something like "this task used these veriables, here are their values at the start" would probably be enough. but i'm totally spitballing; no idea if that's feasible. | 17:40 |
| dmsimard | corvus: unless upstream changed things, callbacks have pretty much access to everything (i.e, all hostvars) .. that includes decrypted contents of an ansible vault | 17:41 |
| corvus | dmsimard: i should be careful what i wish for :) | 17:41 |
| dmsimard | oh, it's scary what you find when you reverse engineer undocumented APIs :D | 17:42 |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul master: Rename javascript package to @zuul-ci/dashboard https://review.openstack.org/551999 | 17:42 |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul master: Reorganize REST API and dashboard urls https://review.openstack.org/556967 | 17:42 |
| dmsimard | Tut really you should be able to get the vars but it's "hidden" in an object like "play.get_variable_manager()._hostvars" | 17:43 |
| dmsimard | s/Tut/But/ | 17:43 |
| dmsimard | just need to know where (or when) you need to get that "snapshot" (task on start, runner_on_ok, etc.) | 17:43 |
| mordred | corvus, tristanC, pabelanger, dmsimard, tobiash: ^^ ok, https:////review.openstack.org/556967 should pass tests now - and I also updated the docs to include info on the rewrite rules | 17:44 |
| dmsimard | Shrews: I was confused about the ZFS part until I understood it wasn't about the filesystem :) | 17:45 |
| clarkb | dmsimard: ya that has confused me too. Not sure I'm a fan of overloading that one :) doesn't help I actually run ZFS at home | 17:46 |
| *** JasonCL_ has quit IRC | 17:48 | |
| Shrews | yeah, well, "zuul from scratch" takes up too much of the 1st line character limit, so we make compromises | 17:48 |
| Shrews | :) | 17:49 |
| mordred | Shrews: maybe we should call it ZfS to differentiate | 17:49 |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul master: Reorganize REST API and dashboard urls https://review.openstack.org/556967 | 17:52 |
| mordred | sorry - left out an update to the html/js side | 17:52 |
| *** JasonCL has joined #zuul | 17:52 | |
| openstackgerrit | Merged openstack-infra/zuul-jobs master: Add a flag to disable no_log on upload-logs https://review.openstack.org/556877 | 17:56 |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul master: Reorganize REST API and dashboard urls https://review.openstack.org/556967 | 17:56 |
| tobiash | mordred: wasn't white label the tenant-less variant? | 18:00 |
| tobiash | https://review.openstack.org/#/c/556967/2/tests/unit/test_web_urls.py | 18:00 |
| mordred | tobiash: yes - those should be the equiv 'rewrite' rules to the ones needed for doing white-label for real | 18:01 |
| tobiash | mordred: ah yes, nevermind | 18:03 |
| mordred | honestly - I think we could remove that /api/ rewrite in that test since our tests don't execute the js so we don't have a sense of whetheror not the status javascript will fetch the json correcty or not | 18:04 |
| mordred | tristanC: I responded to your comment on https://review.openstack.org/#/c/556967 | 18:08 |
| *** JasonCL has quit IRC | 18:23 | |
| mordred | tobiash: ah - I just realized I can add a test to make sure the js is doing the right thing | 18:23 |
| *** JasonCL has joined #zuul | 18:24 | |
| *** JasonCL has quit IRC | 18:26 | |
| dmsimard | tristanC and I were discussing something this morning and I was wondering if it's a use case that would make sense... | 18:28 |
| dmsimard | right now the sf.io Zuul is reporting to the openstack gerrit as "Software Factory CI" and the RDO Zuul is reporting as "RDO CI" -- I believe that's configured at the gerrit connection level | 18:28 |
| *** JasonCL_ has joined #zuul | 18:29 | |
| dmsimard | We're exploring the possibility of moving/migrating the Zuul things to sf.io (while keeping the review.rdoproject.org gerrit). Does that mean we can no longer report as a distinct gerrit user despite being in a different Zuul tenant ? | 18:29 |
| dmsimard | It's not a deal breaker for us but it feels like a gap | 18:30 |
| clarkb | dmsimard: you could probably have two different connections to the same review.o.o? | 18:30 |
| dmsimard | clarkb: yeah but does that mean processing/scanning the same changes twice ? | 18:31 |
| clarkb | though I guess you may hvae to do funny pipeline configs in that case | 18:31 |
| clarkb | dmsimard: potentially. It will depend on your pipeline configs | 18:31 |
| LinuxJedi | corvus: not sure it helped much unfortunately: http://paste.openstack.org/show/715445/ | 18:31 |
| *** JasonCL_ has quit IRC | 18:33 | |
| *** chrnils has quit IRC | 18:39 | |
| openstackgerrit | David Moreau Simard proposed openstack-infra/zuul-jobs master: Do not merge: Test lack of ipv4 with limestone nodes https://review.openstack.org/557007 | 18:49 |
| *** flepied has quit IRC | 18:51 | |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul master: Use requests instead of urllib.request in tests https://review.openstack.org/554057 | 18:51 |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul master: Reorganize REST API and dashboard urls https://review.openstack.org/556967 | 18:51 |
| mordred | corvus, tristanC, tobiash, dmsimard, pabelanger: ^^ that should pass tests now - and I added a test to show white-labeled apis working as expected (which actually showed a bug, so yay) | 18:53 |
| mordred | I reparented it on top of the 'use requests in tests' patch - mostly because that patch makes it a bit more clearer which thing a test is fetching, so I hope it makes the changes to the tests clearer in terms of what's changing | 18:54 |
| corvus | LinuxJedi: the contents of job-output.json may help -- it should have all the info about the upload-logs role; can you take a look at that (and pastebin it if you can)? | 18:54 |
| dmsimard | Is that a legit error? http://logs.openstack.org/84/556784/2/check/tox-py35-on-zuul/5b9de02/ara/result/5dacbf21-f926-46fb-bbe8-5fd0313d8f17/ | 18:55 |
| dmsimard | zuul unit tests | 18:55 |
| pabelanger | it is a failure, but not sure if seen before | 18:57 |
| pabelanger | failing http://logs.openstack.org/88/556988/1/check/zuul-tox-py35/41b902d/testr_results.html.gz too | 18:58 |
| Shrews | yeah, i think that test has been flapping for some reason | 18:58 |
| LinuxJedi | corvus: http://paste.openstack.org/show/715449/ | 18:58 |
| mordred | LinuxJedi: so - I think we're realy close ... | 19:00 |
| mordred | LinuxJedi: it looks like it copied things to "/srv/static/logs/02/202/2ecebd6fec2981fb32551fd367f11382a9e33b8a/check/scanbuild/091a183/job-output.json.gz | 19:00 |
| mordred | LinuxJedi: do you see that file on your server? | 19:01 |
| LinuxJedi | nope, I created /src/static/logs and set it world writable on both builder and master. No subdirectories were created | 19:01 |
| LinuxJedi | on either machine | 19:01 |
| mordred | *awesome* | 19:01 |
| LinuxJedi | exactly :) | 19:02 |
| corvus | it should be happening on the host where zuul-executor is running | 19:02 |
| openstackgerrit | Matthew Thode proposed openstack-infra/nodepool master: add gentoo systemd image for testing https://review.openstack.org/556661 | 19:02 |
| LinuxJedi | should be, but whatever shouldn't be happening isn't :) | 19:03 |
| pabelanger | corvus: LinuxJedi: mordred: did we allow bindmount to host OS for /src/static so it can get out of bwrap? | 19:03 |
| corvus | yeah, just meant to clarify on which host we should expect to find the file | 19:03 |
| corvus | pabelanger: it should be in the trusted context | 19:04 |
| corvus | which still uses bwrap, but i thought we mounted everything | 19:04 |
| openstackgerrit | Matthew Thode proposed openstack-infra/nodepool master: DNM: testing gentoo nodepool-dsvm https://review.openstack.org/556679 | 19:04 |
| corvus | pabelanger: but that's a good thing to double check | 19:04 |
| mordred | that's a great thing to check though ... | 19:04 |
| mordred | yah | 19:04 |
| corvus | (the last time we did a local copy, that wasn't the case) | 19:04 |
| pabelanger | trusted_rw_paths = ['/srv/static'] I think in zuul.conf | 19:05 |
| mordred | what's the / for the bubblewrap invocatio if we didn't bind-mount it? | 19:06 |
| pabelanger | sorry, trusted_rw_paths=/srv/static for zuul.conf | 19:07 |
| mordred | do we set /tmp/091a183e9e1448c78146dda9dc041e7d as the / ? -if so, one could look for /tmp/091a183e9e1448c78146dda9dc041e7d/srv/static and see if that's where it got copied instead | 19:07 |
| pabelanger | that was puppet syntax we have in system-config | 19:07 |
| pabelanger | i didn't think we mounted / by default in bwrap | 19:08 |
| corvus | it may well be a completely transient directory | 19:10 |
| corvus | though i wonder how creating srv/static worked... | 19:10 |
| corvus | LinuxJedi: are you running zuul-executor as root? | 19:10 |
| LinuxJedi | corvus: no, user is "zuul" | 19:11 |
| mordred | corvus: maybe it created it in the bwrap transient dir and it then went away? | 19:11 |
| corvus | good | 19:11 |
| corvus | mordred: yeah. i'd still expect whatever / is to be read-only | 19:11 |
| mordred | yah | 19:11 |
| corvus | i can investigate that in a bit | 19:11 |
| LinuxJedi | sorry for breaking your toys in weird ways guys :) | 19:11 |
| mordred | LinuxJedi: no, this is great! | 19:12 |
| mordred | LinuxJedi: thanks for working though it with us | 19:12 |
| pabelanger | mordred: yah, my understanding, that would make sense if was inside bwrap then went away | 19:12 |
| corvus | LinuxJedi: while we continue to try to figure out what's up, why don't we assume pabelanger's hunch is right, and add 'trusted_rw_paths=/srv/static' to the [executor] section of zuul.conf, and restart the executor | 19:12 |
| LinuxJedi | sure thing | 19:12 |
| mordred | LinuxJedi: also ... | 19:12 |
| pabelanger | you'll also need to make sure it has the write directory permissions for zuul user too | 19:12 |
| mordred | you should update the post playbook corvus had you make and replace logs.openstack.org with whatever the url prefix should be for the webserver serving your /srv/static dir | 19:13 |
| LinuxJedi | pabelanger: yea, I set it to 777 for now (I know, bad, but I'm just hacking a demo for now) | 19:13 |
| pabelanger | LinuxJedi: don't worry, we did the same with zuulv3-dev IIRC | 19:13 |
| LinuxJedi | mordred: yep, I was going to do that once I had a web server there, after I knew the log copy was working ;) | 19:13 |
| mordred | LinuxJedi: cool. just making sure :) | 19:14 |
| *** JasonCL has joined #zuul | 19:15 | |
| LinuxJedi | there are some Ansible specific stuff I'd like to ask later. Such as how to make the test fail on the first bash failure. I suspect that is just my dumb usage of Ansible though | 19:15 |
| LinuxJedi | woot! That fixed it! Nice one pabelanger :) | 19:15 |
| pabelanger | Yay | 19:16 |
| mordred | LinuxJedi: \o/ | 19:16 |
| *** JasonCL has quit IRC | 19:16 | |
| pabelanger | I can push up a patch for zfs shortly | 19:17 |
| *** JasonCL has joined #zuul | 19:17 | |
| *** elyezer has quit IRC | 19:18 | |
| LinuxJedi | corvus: I've left a bunch of notes on zuul-from-scratch for you (prefixed with LXJ). Hope it helps. | 19:18 |
| LinuxJedi | I mean https://etherpad.openstack.org/p/zuul-quickstart-and-zfs(zuul-from-scratch) | 19:19 |
| corvus | LinuxJedi: cool thanks! you're like, maybe, the third person who isn't the author to run through it | 19:20 |
| Shrews | i told him he was the first. i don't know of 2 others | 19:21 |
| corvus | oh, er, they're all first! | 19:24 |
| corvus | i just confirmed we mount /afs in trusted on our executors for basically the same reason; so it seems very likely we should have expected this to be the solution :) | 19:25 |
| *** hasharDockerized is now known as hashar | 19:25 | |
| corvus | so when we add this role to zuul-base-jobs, we should document that, and also in ZfS | 19:25 |
| pabelanger | yah, will be helpful for single node deployments | 19:26 |
| *** JasonCL has quit IRC | 19:28 | |
| *** JasonCL has joined #zuul | 19:29 | |
| *** elyezer has joined #zuul | 19:30 | |
| *** JasonCL has quit IRC | 19:30 | |
| *** JasonCL has joined #zuul | 19:32 | |
| *** JasonCL has quit IRC | 19:33 | |
| *** JasonCL has joined #zuul | 19:34 | |
| *** harlowja has joined #zuul | 19:34 | |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul master: Reorganize REST API and dashboard urls https://review.openstack.org/556967 | 19:35 |
| *** JasonCL has quit IRC | 19:38 | |
| openstackgerrit | Merged openstack-infra/zuul-jobs master: Handle blank public_ipv4 https://review.openstack.org/556784 | 19:46 |
| *** JasonCL has joined #zuul | 19:52 | |
| *** flepied has joined #zuul | 19:53 | |
| *** flepied has quit IRC | 19:56 | |
| gundalow | What's the full path to {{ zuul.project.src_dir }}? "{{zuul.executor.work_root }}/{{ zuul.project.src_dir }}" doesn't exist | 19:56 |
| *** flepied has joined #zuul | 19:56 | |
| *** JasonCL has quit IRC | 19:59 | |
| *** flepied has quit IRC | 20:02 | |
| *** flepied has joined #zuul | 20:03 | |
| mordred | gundalow: zuul.project.src_dir will be on the remote node - and it is, by default, /home/zuul/src/{{ zuul.project.canonical_name }} ... so for ansible would be /home/zuul/src/github.com/ansible/ansible (for instance) | 20:03 |
| mordred | gundalow: zuul.executor.work_root is on the executor where ansible is being run from | 20:03 |
| gundalow | aah, that makes sense | 20:03 |
| mordred | gundalow: actually - I slightly misspoke | 20:03 |
| mordred | gundalow: zuul.project.src_dir will be "src/github.com/ansible/ansible" | 20:04 |
| gundalow | I wondered if that was the case, but I couldn't do delegate_to: localhost (rightly so), so I couldn't check | 20:04 |
| *** flepied has quit IRC | 20:04 | |
| mordred | it's assumed to be used relative to the remote user home dir | 20:04 |
| corvus | yeah, so it should be usable on either the executor or the remote node | 20:04 |
| corvus | depending on how you prefix it | 20:04 |
| *** flepied has joined #zuul | 20:04 | |
| corvus | gundalow's example is the right way to find it on the executor; mordred's is the right way on the remote node (or, perhaps better would be "{{ ansible_user_dir }}/{{ zuul.project.canonical_name }}") | 20:05 |
| corvus | grr | 20:05 |
| *** flepied has quit IRC | 20:05 | |
| corvus | gundalow's example is the right way to find it on the executor; mordred's is the right way on the remote node (or, perhaps better would be "{{ ansible_user_dir }}/{{ zuul.project.src_dir }}") | 20:05 |
| corvus | did i get that right that time? :) | 20:06 |
| gundalow | remote node is what I'm after | 20:06 |
| gundalow | full path | 20:06 |
| *** flepied has joined #zuul | 20:06 | |
| pabelanger | the info is also logged into zuul-info/inventory.yaml on the logs server, if you want to see the actually variable we setup too | 20:07 |
| gundalow | "/home/zuul/src/{{ zuul.project.src_dir }}" | 20:07 |
| corvus | using ansible_user_dir just makes it a bit more generic/future_proof | 20:07 |
| pabelanger | I want to say sf.io doesn't use /home/zuul, let me check | 20:07 |
| corvus | gundalow: zuul.project.src_dir includes src/ so you can drop that component | 20:07 |
| gundalow | I was using https://softwarefactory-project.io/logs/73/11673/1/gate/sf-rpm-build/3d8904a/zuul-info/inventory.yaml (from someone elses log) I guess I should dump all vars myself to be sure what I'm using | 20:07 |
| gundalow | ack | 20:07 |
| *** JasonCL has joined #zuul | 20:07 | |
| gundalow | Thanks :) | 20:07 |
| pabelanger | yah, /home/zuul-worker is softwarefactory | 20:08 |
| gundalow | sweet | 20:08 |
| gundalow | thanks :) | 20:08 |
| mordred | corvus: https://review.openstack.org/#/c/556967/ is ready for review (as are the two patches it's stacked on) | 20:09 |
| mordred | pabelanger, dmsimard, tristanC: ^^ if you have a sec | 20:10 |
| *** flepied has quit IRC | 20:10 | |
| *** flepied has joined #zuul | 20:11 | |
| *** JasonCL has quit IRC | 20:12 | |
| *** flepied has quit IRC | 20:12 | |
| *** flepied has joined #zuul | 20:13 | |
| pabelanger | mordred: +2 | 20:14 |
| mordred | pabelanger: woot. thanks | 20:14 |
| *** flepied has quit IRC | 20:14 | |
| *** flepied has joined #zuul | 20:15 | |
| corvus | mordred: +2; should we wait for an ack from tristanC? | 20:15 |
| *** flepied has quit IRC | 20:16 | |
| *** flepied has joined #zuul | 20:18 | |
| mordred | corvus: maybe? he left a few inline comments before and didnt' seem worried about the direction - we've got some time while we figure out that puppet failure anyway though | 20:19 |
| *** JasonCL has joined #zuul | 20:24 | |
| *** tosky has quit IRC | 20:25 | |
| *** JasonCL has quit IRC | 20:26 | |
| *** tosky has joined #zuul | 20:30 | |
| openstackgerrit | Merged openstack-infra/zuul master: Rename javascript package to @zuul-ci/dashboard https://review.openstack.org/551999 | 20:34 |
| mordred | corvus: got a sec to look at https://review.openstack.org/#/c/554057/ while I'm poking at the puppet changes? | 20:38 |
| *** JasonCL has joined #zuul | 20:39 | |
| corvus | mordred: lgtm | 20:40 |
| mordred | corvus: sweet, thanks | 20:42 |
| *** JasonCL has quit IRC | 20:42 | |
| openstackgerrit | Matthew Thode proposed openstack-infra/nodepool master: add gentoo systemd image for testing https://review.openstack.org/556661 | 20:43 |
| *** JasonCL has joined #zuul | 20:43 | |
| *** hashar has quit IRC | 20:46 | |
| *** JasonCL has quit IRC | 20:48 | |
| *** acozine1 has quit IRC | 21:09 | |
| *** dkranz has quit IRC | 21:13 | |
| pabelanger | tobiash: I think i am having an issue with plugin detecting with zuul on untrusted jobs, I'm struggling to understand why https://github.com/ansible-network/network-engine/pull/11/files is failing. As I understand ansible-playbook will only load plugins next to a playbook, but since the playbook is under test, there is no plugin, just the top-level directory. Don't know why it is failing | 21:34 |
| pabelanger | http://docs.ansible.com/ansible/devel/plugins/action.html#enabling-action-plugins | 21:35 |
| pabelanger | I am guessing maybe because we are placing the project into the role path by default for zuul-executor, even if job isn't using it? | 21:37 |
| pabelanger | actually, I should be able to test this with a unit test in zuul to better understand | 21:37 |
| *** JasonCL has joined #zuul | 21:39 | |
| *** JasonCL has quit IRC | 21:42 | |
| pabelanger | tobiash: actually, we have something in tree, but not tested. This layout, if contained a zuul.yaml, I would expect to be safe: http://git.openstack.org/cgit/openstack-infra/zuul/tree/tests/fixtures/config/speculative-plugins/git/org_project3 | 21:51 |
| pabelanger | since the play is one folder lower then filter_plugins | 21:51 |
| pabelanger | but, I think because we load it into role path, it is failing | 21:52 |
| pabelanger | need to grab some food | 21:53 |
| *** JasonCL has joined #zuul | 21:54 | |
| *** JasonCL has quit IRC | 21:56 | |
| *** ChanServ changes topic to "Discussion of the project gating system Zuul | Docs: https://zuul-ci.org/docs/ | Source: https://git.openstack.org/cgit/openstack-infra/zuul/ | Roadmap: https://storyboard.openstack.org/#!/board/53 | Channel logs: http://eavesdrop.openstack.org/irclogs/%23zuul/" | 21:57 | |
| corvus | Shrews: topic changed :) | 21:57 |
| *** JasonCL has joined #zuul | 21:57 | |
| *** JasonCL has quit IRC | 22:02 | |
| pabelanger | corvus: tobiash: okay, I think I see what is happening, http://git.openstack.org/cgit/openstack-infra/zuul/tree/zuul/executor/server.py#n1232 is what is adding the project into the role path, which is then failing on _blockPluginDirs(). maybe we could discuss at once 3.0 is released out projects could maybe opt out of being added to the role path on the executor | 22:11 |
| pabelanger | how* projects | 22:11 |
| corvus | pabelanger: we could probably just have it automatically not add the project to the role path if it hits the plugin block | 22:14 |
| pabelanger | corvus: yah, that would actually work here | 22:16 |
| *** yolanda has quit IRC | 22:25 | |
| *** yolanda has joined #zuul | 22:27 | |
| *** JasonCL has joined #zuul | 22:29 | |
| *** elyezer has quit IRC | 22:31 | |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul master: Update docs with better websocket instructions https://review.openstack.org/557083 | 22:32 |
| *** JasonCL has quit IRC | 22:33 | |
| openstackgerrit | Matthew Thode proposed openstack-infra/nodepool master: DNM: testing gentoo nodepool-dsvm https://review.openstack.org/556679 | 22:40 |
| *** elyezer has joined #zuul | 22:41 | |
| pabelanger | corvus: fungi: what about testing SetEnvIf and check remote_host first to be more explicit | 22:45 |
| corvus | pabelanger: wrong channel? | 22:45 |
| pabelanger | sorry | 22:45 |
| *** jimi|ansible has quit IRC | 22:58 | |
| *** JasonCL has joined #zuul | 23:03 | |
| *** JasonCL has quit IRC | 23:04 | |
| *** tosky has quit IRC | 23:25 | |
| *** harlowja has quit IRC | 23:32 | |
| *** JasonCL has joined #zuul | 23:32 | |
| *** JasonCL has quit IRC | 23:36 | |
| *** jimi|ansible has joined #zuul | 23:53 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!