| *** mattw4 has quit IRC | 00:00 | |
| *** igordc has quit IRC | 00:10 | |
| *** jamesmcarthur has joined #zuul | 00:21 | |
| *** jamesmcarthur has quit IRC | 00:30 | |
| openstackgerrit | Kerby proposed zuul/nodepool master: AWS driver: add ability to determine AMI id using filters https://review.opendev.org/683205 | 00:37 |
|---|---|---|
| openstackgerrit | Kerby proposed zuul/nodepool master: AWS driver: add ability to determine AMI id using filters https://review.opendev.org/683205 | 00:52 |
| *** swest has quit IRC | 01:35 | |
| *** swest has joined #zuul | 01:51 | |
| *** jamesmcarthur has joined #zuul | 01:54 | |
| *** jamesmcarthur has quit IRC | 02:15 | |
| *** noorul has joined #zuul | 02:18 | |
| *** jamesmcarthur has joined #zuul | 02:26 | |
| *** roman_g has quit IRC | 02:34 | |
| noorul | How do I specify multi tenant configuration in zuul main.yaml? | 02:36 |
| *** rlandy|bbl is now known as rlandy | 02:37 | |
| *** bhavikdbavishi has joined #zuul | 02:49 | |
| fungi | noorul: just add another - tenant: to the top level list, see https://opendev.org/openstack/project-config/src/branch/master/zuul/main.yaml for an example | 02:52 |
| *** bhavikdbavishi1 has joined #zuul | 02:52 | |
| *** jamesmcarthur has quit IRC | 02:52 | |
| fungi | that file has 4 tenants defined currently | 02:52 |
| *** jamesmcarthur has joined #zuul | 02:54 | |
| *** bhavikdbavishi has quit IRC | 02:54 | |
| *** bhavikdbavishi1 is now known as bhavikdbavishi | 02:54 | |
| noorul | fungi: Thank you | 02:54 |
| noorul | fungi: Is it possible to specify cross tenant dependency in the job? | 02:54 |
| noorul | fungi: I mean required-projects attribute of job | 02:55 |
| fungi | builds don't cross tenants, so i'm not sure what you mean. a tenant can include the same repositories and job definitions as another tenant if that's what you're asking | 02:56 |
| fungi | or stated differently, a repository can be present in multiple tenants | 02:56 |
| fungi | though it's strongly suggested that a given project only be gated by one tenant, even if it appears in multiple tenants | 02:57 |
| fungi | having pipelines in more than one tenant fight over which one should merge changes for the same project would get messy | 02:57 |
| noorul | fungi: I agree | 02:58 |
| noorul | but in our organization there is a separation of repos by projects on bitbucket | 02:59 |
| noorul | Say, project1 has p1repo1 p1repo2 p1repo3 | 02:59 |
| noorul | project2 has p2repo1 p2repo2 p2repo3 | 02:59 |
| noorul | So, if I define two tenants for each project as project1 and project2 | 03:00 |
| noorul | Can a job defined in p1repo1 use p2prepo1 in required-projects ? | 03:01 |
| fungi | yes, but to do that you need to add p2prepo1 to the list of projects in your project1 tenant | 03:02 |
| fungi | (terminology is a little confusing, but it sounds like you're mapping bitbucket repos to zuul projects, and bitbucket projects to zuul tenants) | 03:02 |
| noorul | Isn't that correct? | 03:03 |
| fungi | i can only assume it's correct for you | 03:03 |
| noorul | What was the reason behind introducing tenant concept? | 03:03 |
| fungi | the ability to have one zuul serve multiple organizations which want to have isolation from each other | 03:04 |
| noorul | Oh I see | 03:05 |
| *** rfolco has quit IRC | 03:05 | |
| noorul | Here we have multiple bitbucket projects for different programs inside the organization | 03:05 |
| noorul | I am trying to map tenant to a program | 03:06 |
| fungi | for example, opendev's zuul deployment has a "zuul" tenant which serves the zuul community and an "openstack" tenant which serves the openstack community. they each have their own project lists, pipelines and so on, but that doesn't mean they can't both use jobs or refs from some of the same repositories | 03:06 |
| noorul | From zuul docs | 03:09 |
| noorul | A project corresponds to a source code repository with which Zuul is configured to interact. | 03:09 |
| fungi | for example, in that main.yaml i linked, the zuul tenant's untrusted projects include openstack/openstacksdk but is "includes" an empty list from that project, indicating it doesn't even load job configuration, but can still be used for depends-on relationships and required-projects entries in jobs used by the zuul tenant. that same openstack/openstacksdk project also appears in the openstack tenant but with no | 03:10 |
| fungi | such restrictions | 03:10 |
| fungi | on the other hand, in the opendev tenant, the zuul/zuul project is listed with include: [job, secret, nodeset] so the opendev tenant will load jobs, secrets and nodesets from that project even though it's also included unrestricted in the zuul tenant | 03:12 |
| fungi | there are quite a few examples in that file of projects appearing in more than one tenant, with different restrictions and for varying reasons | 03:13 |
| noorul | Too many things to learn | 03:14 |
| noorul | If I use project using include, will it be gated together? | 03:18 |
| noorul | I mean will depends-on work? | 03:18 |
| *** jamesmcarthur has quit IRC | 03:18 | |
| *** jamesmcarthur has joined #zuul | 03:18 | |
| fungi | yes, if all you want is to be able to use depends-on and required-projects with a project which is being gated in a different tenant, the include with the [] (empty list) should be sufficient | 03:19 |
| fungi | and will still prevent that tenant from loading any job, role or nodeset definitions from the project | 03:19 |
| *** jamesmcarthur has quit IRC | 03:22 | |
| fungi | and yeah, it's definitely more to learn, but you only really need to know about it if you're going to be running with multiple tenants and trying to share things between them | 03:22 |
| *** jamesmcarthur has joined #zuul | 03:22 | |
| fungi | i wouldn't personally suggest multiple tenants unless you really need significant isolation between them though. you can still rely on zuul's usual per-project keying and project self-control to allow different teams in an organization to collaborate across projects while retain control of their projects | 03:24 |
| fungi | as long as sharing a common set of trusted/config repositories among them is acceptable | 03:25 |
| fungi | (for example, you could put your zuul administrators in control of the config repository for the tenant, but still allow departments to put job configuration of their own in the projects they control) | 03:26 |
| *** jamesmcarthur has quit IRC | 03:26 | |
| fungi | the opendev deployment only recently became multi-tenant, prior to that we had ~2k projects managed by possibly hundreds of teams all coexisting happily in a single tenant | 03:27 |
| *** noorul has quit IRC | 03:27 | |
| *** noorul has joined #zuul | 03:28 | |
| fungi | the main thing which drove us to wanting additional tenants was so we could define separate sets of pipelines with different kinds of rules specific to each tenant | 03:28 |
| noorul | Thank you! | 03:29 |
| noorul | I think I should go with one tenant, even though there are two separate projects in bitbucket | 03:31 |
| noorul | It it will make it simple | 03:31 |
| noorul | Maybe later just in case required can be moved out | 03:31 |
| *** rlandy has quit IRC | 03:38 | |
| fungi | sounds like a decent plan | 03:41 |
| *** jamesmcarthur has joined #zuul | 04:10 | |
| *** swest has quit IRC | 04:32 | |
| *** jamesmcarthur has quit IRC | 04:41 | |
| *** noorul has quit IRC | 05:16 | |
| *** noorul has joined #zuul | 05:16 | |
| *** swest has joined #zuul | 05:22 | |
| *** pcaruana has joined #zuul | 05:28 | |
| noorul | mordred shared http://paste.openstack.org/show/777615/ example. But for me it fails with | 05:31 |
| noorul | extra keys not allowed @ data['branch'] | 05:31 |
| noorul | fungi: ^^ | 05:40 |
| noorul | Oh document says it is branches | 05:50 |
| *** pcaruana has quit IRC | 05:53 | |
| *** gtema has joined #zuul | 06:03 | |
| *** avass has joined #zuul | 06:04 | |
| *** gtema has quit IRC | 06:11 | |
| *** zbr is now known as zbr|ruck | 06:48 | |
| *** tosky has joined #zuul | 07:18 | |
| *** yolanda has quit IRC | 07:21 | |
| *** yolanda has joined #zuul | 07:23 | |
| *** yolanda has quit IRC | 07:32 | |
| *** hashar has joined #zuul | 07:34 | |
| *** jpena|off is now known as jpena | 07:42 | |
| *** jamesmcarthur has joined #zuul | 07:43 | |
| *** jamesmcarthur has quit IRC | 07:47 | |
| *** yolanda has joined #zuul | 07:47 | |
| *** sileht has quit IRC | 07:49 | |
| *** sileht has joined #zuul | 07:50 | |
| *** sileht has quit IRC | 07:51 | |
| *** sileht has joined #zuul | 07:51 | |
| *** noorul has quit IRC | 08:25 | |
| *** noorul has joined #zuul | 08:26 | |
| *** noorul has quit IRC | 08:32 | |
| *** noorul has joined #zuul | 08:35 | |
| *** roman_g has joined #zuul | 08:44 | |
| *** pcaruana has joined #zuul | 09:12 | |
| *** igordc has joined #zuul | 09:15 | |
| *** igordc has quit IRC | 09:20 | |
| *** igordc has joined #zuul | 09:39 | |
| *** gtema_ has joined #zuul | 09:43 | |
| *** sean-k-mooney has quit IRC | 09:57 | |
| *** bhavikdbavishi has quit IRC | 09:57 | |
| recheck | [so-zuul] Is there any relationship between Zuul and tox for openstack project? → https://stackoverflow.com/questions/58026167/is-there-any-relationship-between-zuul-and-tox-for-openstack-project | 10:10 |
| *** hashar has quit IRC | 10:28 | |
| *** pcaruana has quit IRC | 10:56 | |
| *** hashar has joined #zuul | 10:56 | |
| *** jpena is now known as jpena|lunch | 11:01 | |
| *** panda is now known as panda|lunch | 11:03 | |
| openstackgerrit | Jan Kubovy proposed zuul/zuul master: Github connection timeout configuration https://review.opendev.org/683366 | 11:24 |
| *** igordc has quit IRC | 11:26 | |
| *** pcaruana has joined #zuul | 11:29 | |
| *** Miouge has quit IRC | 11:29 | |
| *** Miouge has joined #zuul | 11:31 | |
| *** chandankumar is now known as raukadah | 11:33 | |
| *** pcaruana has quit IRC | 11:42 | |
| *** pcaruana has joined #zuul | 11:42 | |
| *** dmellado has quit IRC | 11:45 | |
| *** dmellado has joined #zuul | 11:48 | |
| *** jpena|lunch is now known as jpena | 11:53 | |
| *** jamesmcarthur has joined #zuul | 12:13 | |
| *** panda|lunch is now known as panda | 12:16 | |
| noorul | Where should I put project specific roles? | 12:18 |
| *** Miouge has quit IRC | 12:19 | |
| *** Miouge has joined #zuul | 12:22 | |
| *** jamesmcarthur has quit IRC | 12:29 | |
| *** jamesmcarthur has joined #zuul | 12:30 | |
| *** jamesmcarthur has quit IRC | 12:35 | |
| *** rlandy has joined #zuul | 12:38 | |
| *** gtema_ has quit IRC | 12:39 | |
| fungi | noorul: they can go in the projects they're specific to, or in a central shared repository if you prefer | 12:39 |
| *** rfolco has joined #zuul | 12:46 | |
| noorul | fungi: Should the roles folder be at the root ? | 12:47 |
| *** fdegir has quit IRC | 12:47 | |
| *** fdegir has joined #zuul | 12:47 | |
| avass | noorul: yep | 12:48 |
| avass | noorul: https://docs.ansible.com/ansible/latest/user_guide/playbooks_reuse_roles.html#role-search-path | 12:49 |
| avass | noorul: actually depends on where you put the playbook files | 12:49 |
| fungi | noorul: also the 4th paragraph of https://zuul-ci.org/docs/zuul/user/config.html#attr-job.roles talks about directory structure for including roles | 12:50 |
| *** jamesmcarthur has joined #zuul | 12:55 | |
| noorul | Thanks | 12:55 |
| *** jamesmcarthur has quit IRC | 13:02 | |
| *** jamesmcarthur has joined #zuul | 13:03 | |
| *** jamesmcarthur has quit IRC | 13:08 | |
| *** jamesmcarthur has joined #zuul | 13:24 | |
| Shrews | oh for crying out loud. https://review.opendev.org/683250 didn't work??? | 13:26 |
| mordred | Shrews: it's because of crazypants | 13:28 |
| Shrews | zuul.GerritConnection: Received: 404 Not found: test1~master~I431c9604ff5f8c3243ed1d2b6fcc7d16a4fb31dc | 13:28 |
| Shrews | gah, same error | 13:28 |
| mordred | Shrews: (honestly, I don't know what's up with that run - I rechecked it - the child patch worked fine) | 13:28 |
| Shrews | mordred: it was supposed to fix that ^^^ | 13:28 |
| mordred | Shrews: yeah? but like - why did the child patch work? | 13:28 |
| Shrews | mordred: i don't see a child | 13:29 |
| mordred | Shrews: https://review.opendev.org/#/c/683006 | 13:29 |
| Shrews | oh | 13:29 |
| Shrews | i was thinking "parent" when reading "child" | 13:30 |
| mordred | Shrews: there's really no difference | 13:30 |
| Shrews | mordred: yeah, i dunno man | 13:30 |
| Shrews | mordred: corvus: ah ha | 13:31 |
| Shrews | release 1.25.5 released 12 hours ago | 13:32 |
| noorul | I am getting the following error | 13:32 |
| noorul | http://paste.openstack.org/show/778045/ | 13:32 |
| Shrews | and that only blocks 1.25.4 | 13:32 |
| noorul | I added a role in the repo and used it in the playbook | 13:34 |
| Shrews | mordred: perhaps we should change that to urllib3<=1.25.3 ? | 13:34 |
| openstackgerrit | David Shrewsbury proposed zuul/zuul master: Don't use urllib3 1.25.4 https://review.opendev.org/683250 | 13:35 |
| avass | Shrews: mordred: That looks similar to a bug I reported for gertty, is it the encoding of '~' characters? | 13:37 |
| fungi | avass: yes | 13:37 |
| Shrews | avass: yes. see the github link in the requirements.txt change | 13:37 |
| avass | Just saw the github link | 13:37 |
| noorul | fungi: http://paste.openstack.org/show/778045/ | 13:40 |
| mordred | Shrews: jeeze | 13:41 |
| noorul | Need help in figuring out http://paste.openstack.org/show/778045/ | 13:51 |
| fungi | noorul: "ERROR! no action detected in task. [...] name: Run test infrastructure" | 13:51 |
| fungi | does that task have an action? | 13:51 |
| noorul | http://paste.openstack.org/show/778046/ | 13:52 |
| noorul | This is test job http://paste.openstack.org/show/778047/ | 13:54 |
| fungi | is "debug" a valid task? | 13:56 |
| fungi | i'm afraid i'm a bit out of my depth there | 13:56 |
| avass | Yep | 13:56 |
| noorul | Example says so: https://docs.ansible.com/ansible/latest/modules/docker_compose_module.html | 13:56 |
| avass | could it be the '---' at the beginning of the file? | 13:57 |
| noorul | Trying without it | 13:58 |
| fungi | that shouldn't matter, i don't think. there's an implicit --- at the start of any yaml document | 13:58 |
| fungi | so including an explicit one results in the same parsed data as far as i'm aware | 13:58 |
| avass | then I would guess the ' name: Run test infrastructure' | 13:58 |
| avass | is the cause | 13:58 |
| fungi | well, not necessarily, ansible's error said "...may be elsewhere in the file depending on the exact syntax problem" | 13:59 |
| fungi | but at any rate, it seems like ansible is complaining that the play isn't syntactically correct | 14:00 |
| noorul | Removing name did not help | 14:01 |
| noorul | This is the latest http://paste.openstack.org/show/778048/ | 14:02 |
| mordred | noorul: on line 17 - are those actually backticks? If so - can you try removing them | 14:03 |
| noorul | trying | 14:04 |
| noorul | No luck :( | 14:06 |
| mordred | hrm | 14:06 |
| *** openstackgerrit has quit IRC | 14:06 | |
| noorul | Let me try the first step | 14:07 |
| fungi | yeah, maybe it can be bisected by reducing the number of tasks initially, or starting with a noop task of some sort | 14:09 |
| noorul | Started with a debug msg | 14:10 |
| noorul | I think there is some other issue | 14:11 |
| noorul | http://paste.openstack.org/show/778051/ | 14:11 |
| noorul | I now have that alone | 14:11 |
| fungi | i wonder if the "did not meet host_list requirements...did not meet script requirements" warnings are related, or expected | 14:12 |
| * noorul is lost | 14:13 | |
| noorul | This is the playbook http://paste.openstack.org/show/778052/ | 14:15 |
| noorul | This is the roles main.yaml | 14:15 |
| noorul | b | 14:15 |
| noorul | http://paste.openstack.org/show/778053/ | 14:15 |
| fungi | https://github.com/ansible/ansible/issues/48859 seems to suggest the "did not meet <foo> requirements" messages are benign noise | 14:16 |
| fungi | that last paste looks like a playbook not a role? | 14:19 |
| fungi | at least i'm not used to seeing a role yaml file contain a hosts: or tasks: key | 14:19 |
| fungi | i thought the top-level object in the yaml file was just a list and each element a task unto itself | 14:20 |
| *** hashar has quit IRC | 14:20 | |
| fungi | but i'll admit i have at best a tenuous grasp of the various ways these things can be defined in ansible | 14:20 |
| *** hashar has joined #zuul | 14:21 | |
| Shrews | why does the role main.yaml specify hosts? i don't think you can do that | 14:22 |
| fungi | yeah, that's basically what i was asking | 14:23 |
| Shrews | noorul: you might find it helpful to first test your roles outside of zuul before deploying them | 14:23 |
| fungi | it looks like a playbook and not a role | 14:23 |
| Shrews | substitute strings for zuul vars, etc | 14:23 |
| Shrews | the role in http://paste.openstack.org/show/778053/ is especially easy to test that way | 14:24 |
| corvus | i looked back across the pastes, and maybe i missed it, but i couldn't find a paste which actually contained the excerpted text from the error. | 14:24 |
| *** noorul has quit IRC | 14:25 | |
| *** noorul has joined #zuul | 14:25 | |
| *** openstackgerrit has joined #zuul | 14:26 | |
| openstackgerrit | James E. Blair proposed zuul/zuul master: Fix gerrit errors from production https://review.opendev.org/683006 | 14:26 |
| fungi | what noorul said initially was that http://paste.openstack.org/show/778045/ was the error and http://paste.openstack.org/show/778046/ the (playbook?) associated with that error | 14:26 |
| corvus | fungi: right. where is "Run test infrastructure" in 778046? | 14:27 |
| fungi | line 3 | 14:27 |
| corvus | fungi: ah thank you :) | 14:27 |
| fungi | but now i'm starting to wonder that playbook-looking file was actually referenced as a role not a playbook | 14:28 |
| noorul | fungi: You are right | 14:28 |
| noorul | corvus: Any idea when bitbucket plugin will get merged? | 14:29 |
| corvus | yes, that file is a playbook, presumably ac/monitoring/playbooks/testjob.yaml but the error it cites is for a role at ac/monitoring/roles/test-infra/tasks/main.yml | 14:29 |
| Shrews | i think that "name: Run test infrastructure" defined below "hosts:" is also wrong | 14:29 |
| corvus | Shrews: playbooks can have names | 14:29 |
| corvus | but playbooks can't be used as roles | 14:29 |
| corvus | Shrews: er, plays can have names | 14:30 |
| corvus | it looks like a valid playbook, but not a valid task list | 14:30 |
| Shrews | i have never seen that | 14:30 |
| fungi | noorul: so it sounds like you need to restructure that main.yaml content as a role... drop the hosts and tasks keys and move the task definitions to be the top-level list elements in that file instead | 14:30 |
| noorul | Now I have this in the role | 14:31 |
| noorul | http://paste.openstack.org/show/778054/ | 14:31 |
| noorul | Fingers crossed | 14:31 |
| fungi | noorul: that looks more like what i would expect in a role, yes | 14:31 |
| noorul | No luck :( | 14:32 |
| noorul | http://paste.openstack.org/show/778055/ | 14:33 |
| corvus | noorul: regarding merging the driver, i don't believe the review comments have been addressed. that needs to happen first. if ofosos isn't able to, you may feel free to do so as followup changes. | 14:33 |
| ofosos | corvus: I was sick the last two weeks, I'll try to make some time this weekend. I wouldn't mind if noorul works on some stuff | 14:34 |
| noorul | ofosos: I can try dependent PR | 14:35 |
| noorul | ofosos: Glad you are back | 14:35 |
| corvus | ofosos: sorry to hear that, hope you're feeling better. your health is more important than the bitbucket driver :) | 14:35 |
| ofosos | And today I spent the day debugging some janky ethernet cabling in our open workshop :( | 14:35 |
| ofosos | That said, I can now sit next to the 3d printer and work on Zuul, without seeing my internet connection drop out every 2 minutes. | 14:36 |
| noorul | I am stuck with this role stuff :( | 14:37 |
| corvus | my desk is also right next to the 3d printer; but it's a small office so i'm next to everything really | 14:37 |
| fungi | noorul: maybe it doesn't recognize the docker_compose task... https://docs.ansible.com/ansible/latest/modules/docker_compose_module.html suggests before ansible 2.8 it was called docker_service instead. what version of ansible is zuul using for that job? | 14:38 |
| ofosos | noorul: the problem with the dependent pipeline not merging properly is this: we fetch a "generation id" from bitbucket for each pr, the we test, merge number one, by passing the generation id. As soon as that happens, the generation of the second pr is invalid, because the target branch changed. Now to make this happen we need to wiggle the fetch of this id closer to the merge action. | 14:39 |
| corvus | noorul, fungi: zuul uses ansible 2.7 by default currently: https://zuul-ci.org/docs/zuul/user/config.html#attr-job.ansible-version | 14:40 |
| ofosos | Ideally it needs to move to the reporter. That would eliminate that problem. | 14:40 |
| ofosos | The problem with the workshop is that, we're kind of small and still a young institution and if you don't repair the cabling yourself, nobody will do it. | 14:41 |
| noorul | Oh I see | 14:41 |
| noorul | corvus: What do you recommend? 2.7 or 2.8 ? | 14:42 |
| noorul | Assuming there could be a reason why 2.8 is not default | 14:42 |
| corvus | noorul: we're just slow about upgrading; use whichever works in your environment (is most compatible with other tooling). | 14:43 |
| fungi | noorul: the proposed change to switch the default to 2.8 is https://review.opendev.org/676695 | 14:43 |
| corvus | pabelanger: ^ i wonder if we should consider adding 'min-ansible-version' for cases like this? so folks could set a minimum and not have to worry about unpinning | 14:44 |
| *** mattw4 has joined #zuul | 14:45 | |
| *** mattw4 has quit IRC | 14:46 | |
| *** mattw4 has joined #zuul | 14:46 | |
| *** mattw4 has quit IRC | 14:48 | |
| *** mattw4 has joined #zuul | 14:48 | |
| corvus | clarkb: when you awaken, can you review https://review.opendev.org/683006 ? i'd really like to restart opendev with that and the autohold stuff today | 14:51 |
| corvus | oh, wait, clarkb is going to a wedding or something | 14:52 |
| openstackgerrit | Merged zuul/zuul master: Don't use urllib3 1.25.4 https://review.opendev.org/683250 | 14:52 |
| corvus | fungi, Shrews: could you take a look at https://review.opendev.org/683006 ? it fixes broken master so is moderately urgent | 14:53 |
| Shrews | yes | 14:54 |
| corvus | i've approved the autohold stack again | 14:55 |
| noorul | Ansible pinning worked | 14:58 |
| noorul | Woo | 14:58 |
| fungi | 683006 lgtm | 14:58 |
| noorul | Thank you all for the support | 14:59 |
| noorul | Zuul team rocks | 14:59 |
| corvus | noorul: \o/ | 14:59 |
| corvus | we now have a 13-deep gate queue... that's like nova territory :) | 14:59 |
| fungi | like nova territory if there were no other openstack projects ;) | 15:02 |
| *** jamesmcarthur has quit IRC | 15:05 | |
| *** jamesmcarthur has joined #zuul | 15:07 | |
| Shrews | if we want to add to it, this nodepool builder fix would be nice to have: https://review.opendev.org/681857 | 15:09 |
| *** bhavikdbavishi has joined #zuul | 15:11 | |
| corvus | Shrews: +2 with comment | 15:21 |
| maxamillion | hell all, random question but is there any type of "visualization" in Zuul that's similar to Jenkins Blue Ocean? | 15:23 |
| *** mattw4 has quit IRC | 15:24 | |
| maxamillion | oh, the Ara Report ... swank | 15:24 |
| maxamillion | nvm me, this is great :) | 15:24 |
| Shrews | corvus: oh, good point | 15:37 |
| openstackgerrit | James E. Blair proposed zuul/zuul-registry master: Add README and license https://review.opendev.org/683426 | 15:38 |
| *** jamesmcarthur has quit IRC | 15:38 | |
| noorul | ofosos: Have you deployed zuul in production? | 15:39 |
| openstackgerrit | David Shrewsbury proposed zuul/nodepool master: Do not overwrite image upload ZK data on delete https://review.opendev.org/681857 | 15:41 |
| openstackgerrit | David Shrewsbury proposed zuul/nodepool master: Do not overwrite image upload ZK data on delete https://review.opendev.org/681857 | 15:43 |
| *** jamesmcarthur has joined #zuul | 15:43 | |
| fungi | maxamillion: also the "console" tab on the build detail pages | 15:43 |
| maxamillion | fungi: +1 thanks | 15:43 |
| *** jamesmcarthur has quit IRC | 15:45 | |
| *** sgw has quit IRC | 15:49 | |
| *** armstrongs has joined #zuul | 15:51 | |
| *** jamesmcarthur has joined #zuul | 15:54 | |
| *** noorul has quit IRC | 15:56 | |
| armstrongs | hey guys, quick question i have an .zuul.yaml file for my ansible role which lives in the directory of my role. The .zuul.yaml is in my ansible-role top level directory structure. I want to spin up a nodepool vm and execute that role on the target nodepool server. What is the best way to set-up the zuul.yaml config to achieve that? How do i tell | 15:56 |
| armstrongs | zuul to execute the current workspaces role? | 15:56 |
| openstackgerrit | Merged zuul/zuul master: Store autohold requests in zookeeper https://review.opendev.org/661114 | 15:58 |
| openstackgerrit | Merged zuul/zuul master: Add caching of autohold requests https://review.opendev.org/663412 | 15:58 |
| *** noorul has joined #zuul | 15:59 | |
| noorul | ofosos: hi | 16:01 |
| *** jamesmcarthur_ has joined #zuul | 16:04 | |
| *** jamesmcarthur has quit IRC | 16:07 | |
| noorul | hostname:port combination in path is creating lot of problem for me | 16:08 |
| corvus | armstrongs: make a playbook in the same repo that invokes your role, then create a zuul job that specifies that playbook as the "run" playbook for the job, then add that job to that project's pipeline config. | 16:10 |
| corvus | armstrongs: everything after writing the playbook happens in your role's .zuul.yaml | 16:11 |
| corvus | armstrongs: the .zuul.yaml for that might look like this: https://etherpad.openstack.org/p/qwWCzT1OJf | 16:12 |
| armstrongs | so the bit i wasnt getting is how zuul knows that the current workspace role is a role in the playbook | 16:12 |
| corvus | armstrongs: ah, good question -- if a project | 16:12 |
| corvus | grr | 16:12 |
| corvus | armstrongs: ah, good question -- if the project looks like it might be an ansible role, or a collection of roles, it automatically adds it to the ansible role path | 16:12 |
| ofosos | noorul: hi | 16:13 |
| corvus | armstrongs: there's a whole bunch of text about it here: https://zuul-ci.org/docs/zuul/user/config.html#attr-job.roles see specifically the last paragraph of that section | 16:13 |
| armstrongs | so i just call in my run playbook the role name of the project i added to untrusted projects and it will link | 16:13 |
| armstrongs | yeah but does that not need to be under repo with a roles folder to work | 16:14 |
| corvus | armstrongs: if it's the *current* project (ie, the project of the change under test) it's automatically added; if it's a different project, list the project under roles (that's the first few paragraphs of that section) | 16:14 |
| corvus | armstrongs: it doesn't have to have a roles dir -- zuul also looks for 'tasks/main.yaml' to see if the project itself is one role | 16:15 |
| corvus | armstrongs: that's described in paragraph 4 :) | 16:15 |
| armstrongs | it is my current project i wana run .zuul.yaml is at the tasks folder level | 16:15 |
| armstrongs | in current workspace/project | 16:15 |
| corvus | you should definitely read the whole section :) | 16:15 |
| armstrongs | yeah haha | 16:15 |
| armstrongs | sorry for being a noob | 16:16 |
| armstrongs | :) | 16:16 |
| corvus | no worries, there's a lot of docs :) | 16:17 |
| fungi | does the ansible 2.5 removal change need advance warning on the ml before we approve it, or has that already happened? presuambly folks will want to know that they need to move any remaining jobs off 2.5 before upgrading | 16:18 |
| *** pcaruana has quit IRC | 16:19 | |
| noorul | Is there any way to remove hostname and port combination from the path | 16:19 |
| noorul | ? | 16:19 |
| noorul | http://paste.openstack.org/show/778141/ | 16:19 |
| noorul | I tried canonical_hostname, but it not only worked, stopped working stuff | 16:20 |
| corvus | noorul: that's how it should be done | 16:21 |
| corvus | so if that didn't work, there are bugs somewhere (perhaps incomplete support in the bb driver?) | 16:21 |
| corvus | fungi: good question -- i think we've done this before, what did we do then? | 16:21 |
| * fungi will research | 16:24 | |
| corvus | fungi: it looks like 91f10d21bdfd09891d382b800debeb8142b605f0 is the last time we dropped support for an ansible version, and that was when we only supported one | 16:25 |
| corvus | so this must be the first time we're dropping a version from a range | 16:25 |
| ofosos | noorul: from which path do you want to remove it? | 16:27 |
| fungi | corvus: yeah, that's what i'm finding too. still, i'll hunt for messaging around that timeframe | 16:27 |
| AJaeger | corvus: want to merge https://review.opendev.org/567696 - "Allow upload-logs to toggle compression of console logs" ? Or better wait? | 16:28 |
| ofosos | noorul: yes, we're running Zuul in production, now deploying 2-3 project using it. | 16:30 |
| corvus | AJaeger: lgtm | 16:30 |
| corvus | fungi: i think i'm happy with the release note in 650431 as messaging. | 16:30 |
| *** jpena is now known as jpena|off | 16:31 | |
| corvus | fungi: if someone wants to send a note to -announce saying "hey we're merging this change, it'll be in the next release" that sounds like a fine idea. but i don't feel it's strictly required. | 16:32 |
| fungi | yeah, i'm not finding any advance warning from when we switched from 2.4 to 2.5. i agree this is slightly different since it's the first time we've removed an ansible since the multi-ansible implementation, but the result is still similar (probably even less impactful) | 16:33 |
| fungi | okay, just wanted to be sure i wasn't jumping the gun of i approved it | 16:33 |
| corvus | Shrews: autohold got this error uploading the image to dockerhub: | 16:36 |
| corvus | 2019-09-20 16:32:23.986559 | ubuntu-bionic | received unexpected HTTP status: 502 Bad Gateway | 16:36 |
| *** sgw has joined #zuul | 16:37 | |
| *** mattw4 has joined #zuul | 16:38 | |
| ofosos | corvus: what's the idea behind 677510 | 16:41 |
| noorul | What is the best practice in placing playbooks folder in a project? | 16:41 |
| ofosos | ? | 16:41 |
| corvus | ofosos: i'm not sure i recall anymore -- i think we were having a conversation and i pushed that up as an illustration of how to convert one of the jobs that's using the fake gerrit to use the fake bitbucket | 16:43 |
| armstrongs | corvus: got that working thanks. Seeing something weird though on the new console output it isn't showing the failure in the logs the failure entry is missing, can see it on ara report and txt output logs though | 16:45 |
| corvus | armstrongs: 2 questions: 1: are you running at least the latest release, and 2: is the log public or can you paste the relevant section of job-output.json? | 16:46 |
| corvus | (we merged some fixes to that recently, but could have missed a case) | 16:47 |
| armstrongs | im running Zuul version: 3.10.2.dev23 b53b6bad | 16:47 |
| armstrongs | i can paste the output of job-output.json | 16:48 |
| armstrongs | 2secs | 16:48 |
| fungi | noorul: jobs refer to playbooks by their path within a repo, so the structure is fairly freeform. in our projects we just create a directory called "playbooks" at the top level of the git repository's worktree and then organize playbooks in individual subdirectories within that, loosely naming the playbook files themselves based on the phases they're called from (pre.yaml, run.yaml...) | 16:48 |
| pabelanger | corvus: maybe, seems like something to explore. So far, each time we (zuul.a.c) have switch to newer version of ansible, this have continued to work. Which avoids the need to have a mix of ansible versions for jobs over trying to keep all jobs the same version. | 16:49 |
| pabelanger | however, I'd just be infavor of upgrading to 2.8 by default | 16:49 |
| pabelanger | given 2.9 is a few weeks away | 16:49 |
| fungi | pabelanger: on a related note, looks like we marked ansible 2.5 as deprecated in https://review.openstack.org/631930 which merged along with the multi-ansible implementation back i march, so that's 6 months... should we be updating 2.6 to say "deprecated" with the removal of 2.5? | 16:49 |
| pabelanger | maxamillion: yah, ara and zuul web is getting better each day | 16:49 |
| pabelanger | fungi: sounds like a good idea | 16:50 |
| pabelanger | can do that when I rebase the review | 16:50 |
| openstackgerrit | Mark Meyer proposed zuul/zuul master: Change reporter pr_version handling https://review.opendev.org/683588 | 16:50 |
| fungi | pabelanger: does it need rebasing? | 16:50 |
| fungi | looked like it was passing | 16:50 |
| pabelanger | 676695 has merge conflict | 16:50 |
| fungi | ahh | 16:51 |
| openstackgerrit | James E. Blair proposed zuul/zuul-jobs master: Retry pushes to dockerhub https://review.opendev.org/683589 | 16:51 |
| corvus | Shrews: ^ | 16:51 |
| pabelanger | same with 674854, can rebase whole stack, adding deprecated info for 2.6 | 16:51 |
| pabelanger | hopefully rc1 is out today | 16:51 |
| pabelanger | will rebase then | 16:51 |
| fungi | awesome | 16:51 |
| ofosos | noorul: do you still have the dependent pipeline setup handy? I would like to test 683588 with this, I think it will make it work. | 16:52 |
| ofosos | Just pushed, wait for the tests... | 16:52 |
| fungi | pabelanger: i think with the discovery that we're marking the oldest supported ansible as "deprecated" in the ansible config, that seems like a nice machine-readable signal deployers can rely on and filter for to identify jobs which need to be switched to newer ansible versions soon | 16:53 |
| openstackgerrit | James E. Blair proposed zuul/zuul-registry master: Add README, license, and Zuul config https://review.opendev.org/683426 | 16:53 |
| ofosos | I was unsure how to handle this, but basically we can reload the pr_version in the reporter. You can configure Bitbucket to remove acknowledgement on a code push and this will still work (as in: no merge) in this case. | 16:53 |
| *** igordc has joined #zuul | 16:55 | |
| *** armstrongs has quit IRC | 16:57 | |
| *** armstrongs has joined #zuul | 16:58 | |
| *** armstrongs has quit IRC | 16:58 | |
| ofosos | Does anybody have the zuul tenant url on zuul.openstack.org I forgot it againt | 16:59 |
| openstackgerrit | Paul Belanger proposed zuul/zuul-jobs master: Allow upload-logs to toggle compression of console logs https://review.opendev.org/567696 | 17:00 |
| *** armstrongs has joined #zuul | 17:00 | |
| pabelanger | fungi: ++ agree | 17:00 |
| fungi | ofosos: you want zuul.opendev.org | 17:01 |
| ofosos | fungi: you're a lifesaver :) | 17:01 |
| fungi | zuul.openstack.org is a whitelabeled view of the openstack tenant on the zuul.opendev.org deployment | 17:01 |
| openstackgerrit | Andreas Jaeger proposed zuul/zuul-jobs master: Allow upload-logs to toggle compression of console logs https://review.opendev.org/567696 | 17:04 |
| pabelanger | corvus: AJaeger: I'm maybe thinking our upload-logs patch for gzip support, should also have a test. But won't get to that today. should have time at fest next week | 17:04 |
| *** hashar has quit IRC | 17:06 | |
| noorul | ofosos: In the middle of migrating a project to zuul | 17:07 |
| openstackgerrit | Merged zuul/nodepool master: Reduce upload threads in tests from 4 to 1 https://review.opendev.org/682977 | 17:09 |
| noorul | ofosos: I am wondering how are tests passing without that change? | 17:09 |
| ofosos | Easy, like I explained: You're merging PR1, that works, then you're merging PR2 with the old 'pr_version' variable which is now outdated (because you merged PR1), and that fails. | 17:10 |
| ofosos | 'pr_version' is a Bitbucket variable, that signifies the PR version for every PR. We have to pull the new 'pr_version' in the reporter and magically everything works. | 17:11 |
| ofosos | That only happens with the dependent pipeline. | 17:11 |
| ofosos | noorul: understood? | 17:12 |
| noorul | ofosos: That is the not the scenario | 17:13 |
| ofosos | Then what's the scenario? | 17:13 |
| noorul | ofosos: PR1 and PR2 approved simultaneously | 17:13 |
| ofosos | Yes? | 17:13 |
| ofosos | We're talking about merging. | 17:13 |
| noorul | ofosos: Both enters the gate queue | 17:13 |
| ofosos | Yes | 17:13 |
| ofosos | And then one after the other gets approved. That is what I'm talking about | 17:14 |
| ofosos | ^ gets merged | 17:14 |
| noorul | ofosos: Actually PR2 has a change on the same line of PR1 | 17:14 |
| noorul | ofosos: PR2 job should not run | 17:14 |
| noorul | That was the scenario I was talking about long back | 17:14 |
| noorul | I haven't re-looked at it lately | 17:15 |
| ofosos | I think we fixed this with changing the branch variable. I'm talking about the problems after that. | 17:15 |
| noorul | I am not able to recollect | 17:16 |
| noorul | I am off to bed now | 17:16 |
| noorul | I will catch you up later | 17:16 |
| ofosos | noorul: good night | 17:16 |
| ofosos | I do have a Bitbucket server in my private lab now and can test. Let's see. I'll probably fire it up tomorrow. | 17:17 |
| openstackgerrit | Paul Belanger proposed zuul/zuul master: Remove support for ansible 2.5 https://review.opendev.org/650431 | 17:18 |
| openstackgerrit | Paul Belanger proposed zuul/zuul master: Switch ansible_default to 2.8 https://review.opendev.org/676695 | 17:18 |
| openstackgerrit | Paul Belanger proposed zuul/zuul master: WIP: Support Ansible 2.9 https://review.opendev.org/674854 | 17:18 |
| Shrews | corvus: i'm beginning to rethink my preference on having mulitple, easily reviewable changes vs. one massive change :/ | 17:20 |
| corvus | Shrews: it's really helping us find bugs | 17:21 |
| corvus | Shrews: if we get really annoyed, we can squash it (review then squash is always an option) | 17:22 |
| *** noorul has quit IRC | 17:22 | |
| openstackgerrit | Paul Belanger proposed zuul/zuul master: Switch ansible_default to 2.8 https://review.opendev.org/676695 | 17:22 |
| openstackgerrit | Paul Belanger proposed zuul/zuul master: WIP: Support Ansible 2.9 https://review.opendev.org/674854 | 17:22 |
| pabelanger | fungi: 2.6 now marked deprecated, with reno notes too | 17:22 |
| fungi | thanks pabelanger! | 17:25 |
| pabelanger | thank you! | 17:25 |
| openstackgerrit | Mark Meyer proposed zuul/zuul master: Change reporter pr_version handling https://review.opendev.org/683588 | 17:26 |
| *** roman_g has quit IRC | 17:31 | |
| *** sgw has quit IRC | 17:33 | |
| corvus | armstrongs sent me some debugging output privately and i think we've come up with a reproducer for that problem | 17:39 |
| corvus | i think https://review.opendev.org/683598 reproduces it -- you can see the output here https://zuul.opendev.org/t/openstack/build/3d35f38b57584ba6adfabcfd5f5230b3/log/job-output.json#4507 and here https://zuul.opendev.org/t/openstack/build/3d35f38b57584ba6adfabcfd5f5230b3/console | 17:39 |
| corvus | but based on https://zuul.opendev.org/t/openstack/build/3d35f38b57584ba6adfabcfd5f5230b3/log/job-output.txt#289 we should see something | 17:40 |
| AJaeger | when do we want to update opendev's instance of Zuul to newer ansible? https://review.opendev.org/#/c/676697/1 proposes to update to 2.7 | 17:40 |
| AJaeger | note that once 676695 merges, we move to 2.8 as well - without 676697 | 17:41 |
| AJaeger | since we use 2.7 already, 676697 looks safe, isn't it? | 17:42 |
| AJaeger | log files show we use 2.7.9 | 17:43 |
| pabelanger | yah, that should be a noop and protects us if we merge the 2.8 bump into zuul (git) and for some reason we restart zuul.o.o. That would switch everything to 2.8 | 17:44 |
| pabelanger | but, i do think we should move opendev to 2.8 | 17:44 |
| pabelanger | however, openstack project likely needs some time? | 17:44 |
| pabelanger | actually, now that I think of it | 17:45 |
| pabelanger | if opendev base job is 2.8, and openstack parent to it, doesn't those jobs also move to 2.8 too? | 17:45 |
| pabelanger | or do we do split ansible-playbook pre run phase | 17:45 |
| pabelanger | doesn't look like it | 17:46 |
| pabelanger | is, in case of opendev, we might need to lock step min version of ansible across all tenants | 17:47 |
| *** sgw has joined #zuul | 17:47 | |
| *** tosky_ has joined #zuul | 17:48 | |
| pabelanger | Hmm, yah, I think we need to test that. Not sure what will happen if we bump opendev tenant, and say openstack stays on 2.7 | 17:49 |
| *** tosky is now known as Guest87259 | 17:51 | |
| *** tosky_ is now known as tosky | 17:51 | |
| AJaeger | interesting, pabelanger - I removed my +2 and wait for tests | 17:51 |
| *** Guest87259 has quit IRC | 17:52 | |
| pabelanger | AJaeger: I could also be wrong :) | 17:52 |
| corvus | pabelanger: where do base jobs come in to play? i thought all the changes were updating default-ansible-version | 17:52 |
| pabelanger | yah, I think 676697 is safe today | 17:52 |
| corvus | (yes, if we set ansible-version on base, it would affect all tenants, but that's not what 676697 does) | 17:52 |
| pabelanger | was saying, if we decide to just upgrade opendev to 2.8, I think we might also need to do all other tenants | 17:53 |
| pabelanger | +1 | 17:53 |
| pabelanger | however | 17:54 |
| pabelanger | it is possible, that is a job is using 2.6 or 2.5 some place, 676697 will upgrade them to 2.7 | 17:54 |
| pabelanger | right? | 17:54 |
| corvus | 697 will only affect jobs which don't specify a version | 17:55 |
| pabelanger | okay cool | 17:55 |
| AJaeger | so, should we merge 697 - or don't need it? | 17:57 |
| *** armstrongs has quit IRC | 18:01 | |
| corvus | AJaeger: it won't be necessary if we merge 676695 instead | 18:07 |
| AJaeger | 695 means we switch to 2.8 - are we ready for that? | 18:08 |
| * AJaeger is not deep enough into ansible to answer that... | 18:08 | |
| corvus | it's past time for zuul to make the switch | 18:09 |
| fungi | yeah, 2.9 is nearly upon us | 18:09 |
| pabelanger | I think 2.8 is stable, but I didn't dig into openstack jobs to ensure things are compat | 18:09 |
| pabelanger | haven't seen any new issues so far | 18:09 |
| Shrews | perhaps we should begin setting the expectation that our version of ansible used in openstack will bump up just after each release of openstack? | 18:16 |
| *** NBorg has joined #zuul | 18:19 | |
| *** igordc has quit IRC | 18:20 | |
| pabelanger | Yah (when working on openstack-infra) we'd try to do that, like image upgrades or job changes at start of cycle. But always a fair bit of work to coordinate that across all projects, in an effort to minimize breakages. For 2.8 switch, tried to ensure zuul tenant was good (after zuul.a.c), but can't give the same guarentee nothing will break for openstack. So, +1 for bump, but unless someone in openstack keeps | 18:21 |
| pabelanger | eye on it (I know most here do) but also a good time to ask the openstask community (or TC) to assist. | 18:21 |
| openstackgerrit | Merged zuul/zuul master: Pagure - handle Pull Request tags (labels) metadata https://review.opendev.org/681050 | 18:22 |
| openstackgerrit | Paul Belanger proposed zuul/zuul master: WIP: Support Ansible 2.9 https://review.opendev.org/674854 | 18:25 |
| pabelanger | updated for ansible 2.9.0rc1 | 18:26 |
| *** igordc has joined #zuul | 18:26 | |
| *** snapiri has quit IRC | 18:26 | |
| *** snapiri has joined #zuul | 18:27 | |
| openstackgerrit | Merged zuul/zuul master: Pagure - reference pipelines add open: True requirement https://review.opendev.org/681252 | 18:29 |
| *** avass is now known as Guest47605 | 18:31 | |
| *** avass has joined #zuul | 18:31 | |
| openstackgerrit | Paul Belanger proposed zuul/zuul master: Remove support for ansible 2.5 https://review.opendev.org/650431 | 18:32 |
| openstackgerrit | Paul Belanger proposed zuul/zuul master: Switch ansible_default to 2.8 https://review.opendev.org/676695 | 18:32 |
| openstackgerrit | Paul Belanger proposed zuul/zuul master: WIP: Support Ansible 2.9 https://review.opendev.org/674854 | 18:32 |
| corvus | mhu isn't here, but tests.unit.test_web.TestTenantScopedWebApi.test_dequeue looks like a flaky test | 18:35 |
| corvus | Shrews: ^ i've seen it fail our changes a couple of times | 18:36 |
| corvus | i think we're going to need to fix that | 18:36 |
| corvus | it has a sleep() in it, which is almost certainly the crux of the problem | 18:36 |
| corvus | i'll take a look after lunch if no one else beats me to it | 18:36 |
| pabelanger | I'm just looking at https://zuul.opendev.org/t/zuul/status and the gate pipeline, and noticed that zuul-upload-image has only run for the first change in the queue. Is there a semaphore on that job, which stops the next patch in the queue from also running it? | 18:38 |
| pabelanger | or, just lacking nodes | 18:38 |
| corvus | pabelanger: no, but there's an artifact dependency (it builds container images). however the second change is just waiting on nodes | 18:38 |
| pabelanger | okay, cool | 18:39 |
| corvus | a potential optimization would be to tell zuul that it's going to rebuild all of the artifacts it depends on, so in this particular case, it doesn't need to wait | 18:39 |
| corvus | (if there were a nodepool or python-base change ahead of it, it would need to wait) | 18:39 |
| pabelanger | k | 18:39 |
| corvus | this is likely to be a common situation, so it's probably worth brainstorming a way to do that | 18:40 |
| corvus | but right now, it just knows that changes ahead of it build container images, and it uses container images | 18:40 |
| pabelanger | okay, understand now. Thanks! | 18:41 |
| corvus | maybe we could do that by saying that the jobs, when run on zuul changes, require "python-base-container-images" and "nodepool-container-images" and they provide "zuul-container-images", and likewise swap that for the jobs when run on nodepool | 18:41 |
| pabelanger | there we go, next change started running zuul-upload-image job | 18:43 |
| corvus | once it pauses, the third can start | 18:43 |
| pabelanger | k | 18:43 |
| pabelanger | maybe I'll look more into this stuff on plane, haven't had chance to use it too much | 18:43 |
| corvus | pabelanger: if you have a quick sec, can you look at https://review.opendev.org/683589 ? | 18:44 |
| pabelanger | sure, just about to relocate home | 18:44 |
| pabelanger | +3 | 18:45 |
| pabelanger | heading out again for few hours! | 18:46 |
| pabelanger | see everything this weekend, going to ansiblefest | 18:46 |
| Shrews | corvus: not even a comment on why that sleep() is there. grrr... | 18:50 |
| gouthamr | hi, quick question, does zuul not persist jobs/results from a dequeued change? | 18:53 |
| gouthamr | i just abandoned a change in openstack/manila, and zuul was done running atleast 2 of the 4 jobs it had for that change, i expected to find the logs in the build history | 18:54 |
| gouthamr | just wondering if this is an incorrect expectation :) | 18:55 |
| Shrews | I'm also leaving a bit early today. Everyone have a good weekend, and safe travels if you're heading out for 'fest. | 18:57 |
| openstackgerrit | Merged zuul/zuul master: Pagure - handles pull-request.closed event https://review.opendev.org/681279 | 19:00 |
| openstackgerrit | Merged zuul/zuul-jobs master: Retry pushes to dockerhub https://review.opendev.org/683589 | 19:02 |
| fungi | gouthamr: the builds are only recorded when their buildset is reported | 19:04 |
| fungi | gouthamr: the data is written to an sql database as a reporter, similar to how it writes results into gerrit comments | 19:04 |
| fungi | so if not all jobs complete, none are recorded | 19:05 |
| *** avass has quit IRC | 19:05 | |
| fungi | there's been some discussion of ways to change that so that builds are recorded into the database individually as they finish, independent of their buildset, but that would be a somewhat significant overhaul | 19:06 |
| *** avass has joined #zuul | 19:06 | |
| fungi | on the other hand, it would allow us to link to build result pages for completed builds in the status display for items whose buildsets are not yet complete, so it would be a compelling improvement | 19:07 |
| gouthamr | oh.. ty fungi - makes sense, there's this cornercase desire i share with TheJulia to see logs from dropped runs for completed jobs, is there any way to retrieve those - they're already in swift somewhere? | 19:08 |
| *** jamesmcarthur_ has quit IRC | 19:10 | |
| *** igordc has quit IRC | 19:13 | |
| fungi | gouthamr: i might be able to find it. what was the change you abandoned? and what pipeline was it being tested in? | 19:13 |
| fungi | though continued discussion of this particular request is probably more appropriate in #openstack-infra | 19:14 |
| gouthamr | fungi: ack, TheJulia hangs out there, so yes... | 19:15 |
| *** bhavikdbavishi has quit IRC | 19:17 | |
| openstackgerrit | Merged zuul/zuul master: Add autohold-info CLI command https://review.opendev.org/662487 | 19:19 |
| *** igordc has joined #zuul | 19:19 | |
| *** jamesmcarthur has joined #zuul | 19:20 | |
| *** tbarron has joined #zuul | 19:20 | |
| *** jamesmcarthur has quit IRC | 19:23 | |
| openstackgerrit | Merged zuul/zuul master: Record held node IDs with autohold request https://review.opendev.org/662498 | 19:48 |
| avass | corvus: so we talked about having multiple reusable workspaces for static nodes when you and mordred were in Gothenburg | 19:50 |
| avass | corvus: looking at the source for static provider it looks like it should be possible to have multiple entries of the same node name with differing users or ports since it identified them as namedtuples | 19:51 |
| avass | corvus: so is there anything else that could require the node names to be unique? not able to test it at the moment | 19:52 |
| avass | corvus: otherwise it looks like the documentation is a bit misleading since the node name+user+port is what needs to be unique | 19:55 |
| *** hashar has joined #zuul | 19:55 | |
| openstackgerrit | Merged zuul/zuul master: Add scheduler config options for hold expiration https://review.opendev.org/682675 | 20:17 |
| corvus | avass: i agree that looks like what the implementation does and the docs are misleading. we should see if there's a functional test that confirms that behavior, and if not, write one, and then update the docs. | 20:20 |
| *** jamesmcarthur has joined #zuul | 20:27 | |
| fungi | zuul-maint: i switched https://storyboard.openstack.org/#!/story/2006526 to public since it's referenced in the release notes for 3.10.2 | 20:27 |
| corvus | fungi: thx | 20:27 |
| *** jamesmcarthur has quit IRC | 20:27 | |
| fungi | i also tagged it zuul-security in case we're still doing that | 20:27 |
| avass | corvus: alright, I think I can look into it a bit more on monday | 20:28 |
| openstackgerrit | Merged zuul/zuul master: Mark nodes as USED when deleting autohold https://review.opendev.org/664060 | 20:37 |
| *** avass has quit IRC | 20:52 | |
| openstackgerrit | Merged zuul/zuul master: Auto-delete expired autohold requests https://review.opendev.org/663762 | 20:59 |
| *** jamesmcarthur has joined #zuul | 21:00 | |
| *** jamesmcarthur has quit IRC | 21:03 | |
| *** rfolco has quit IRC | 21:08 | |
| openstackgerrit | Merged zuul/zuul master: Add autohold delete/info commands to web API https://review.opendev.org/679057 | 21:20 |
| *** rlandy has quit IRC | 21:25 | |
| NBorg | Is there a way to generate jobs for a pipeline dynamically? An easy workaround is to write a script that generates the jobs and the pipeline, but it is always nice to not have to commit generated content. | 21:33 |
| openstackgerrit | Paul Belanger proposed zuul/zuul master: WIP: Support Ansible 2.9 https://review.opendev.org/674854 | 21:39 |
| SpamapS | NBorg: it's not the first time that has been requested, but no. | 21:39 |
| openstackgerrit | Merged zuul/zuul master: Remove outdated TODO https://review.opendev.org/682421 | 21:43 |
| *** hashar has quit IRC | 21:46 | |
| daniel2 | Why did all openstack projects move from github to opendev? | 21:56 |
| SpamapS | they were never intentionally on github. ;) | 21:58 |
| NBorg | SpamapS: Ok, then I'm not missing something in the docs. Thanks. | 21:58 |
| SpamapS | daniel2: https://mako.cc/writing/hill-free_tools.html <-- but otherwise, this. | 21:59 |
| daniel2 | ah. I'm curious if it's possible to mirror the gitea repos in a private gitlab instance, or another gitea instance. We used to have a local mirror of the github openstack repos to improve speed when building images. | 22:00 |
| SpamapS | The canonical home for the code is review.opendev.org, not opendev.org | 22:04 |
| fungi | daniel2: should be just as easy to mirror from gitea as github. it's all git protocol anyway | 22:04 |
| SpamapS | and yeah, same thing really. | 22:05 |
| fungi | daniel2: also, the official openstack repos (the ones you find in https://opendev.org/openstack/ that is) are still being mirrored to https://github.com/openstack/ anyway | 22:06 |
| fungi | and some projects who moved out of the openstack namespace are mirroring their repositories to various orgs on github, but some aren't (notably, the things in https://opendev.org/zuul/ for example) | 22:07 |
| fungi | what's temporarily confusing is some of the stuff which moved out of the openstack namespace in opendev hasn't had its corresponding dead github mirror copies clearly archived | 22:13 |
| tristanC | softwarefactory-project.io's Zuul now has 1775 projects in its tenants configuration :-) | 22:18 |
| fungi | that's basically the same order of magnitude as opendev (2132 at present count, not including gerrit built-in repos) | 22:20 |
| tristanC | fungi: though http://zuul.opendev.org/tenants lists less | 22:21 |
| fungi | good point. we have a bunch of retired repos in opendev | 22:22 |
| fungi | i was looking at gerrit ls-projects | 22:22 |
| SpamapS | tristanC: nice | 22:23 |
| fungi | so not counting opendev's retired repos, sf is larger! | 22:23 |
| tristanC | well, our zuul does include quite a few project's from opendev.org for rdo's third party ci too | 22:24 |
| openstackgerrit | Merged zuul/zuul master: Fix gerrit errors from production https://review.opendev.org/683006 | 22:24 |
| fungi | also i wonder if that double-counts repos presented into more than one tenant | 22:25 |
| tristanC | fungi: tenant's list does include duplicate across tenant | 22:25 |
| tristanC | for good measure, here are the telegraf metrics of the host running the zuul scheduler: https://softwarefactory-project.io/grafana/?orgId=1&var-datasource=default&var-server=zs.softwarefactory-project.io&var-inter=$__auto_interval_inter | 22:27 |
| *** mgoddard has quit IRC | 23:17 | |
| *** mgoddard has joined #zuul | 23:19 | |
| *** tosky has quit IRC | 23:28 | |
| *** mattw4 has quit IRC | 23:48 | |
| *** rfolco has joined #zuul | 23:57 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!