| -@gerrit:opendev.org- Zuul merged on behalf of Simon Westphahl: [zuul/zuul] 859066: Link span of queue item to trigger event span https://review.opendev.org/c/zuul/zuul/+/859066 | 03:33 | |
| -@gerrit:opendev.org- Zuul merged on behalf of Simon Westphahl: | 03:34 | |
| - [zuul/zuul] 859067: Trace received Github events https://review.opendev.org/c/zuul/zuul/+/859067 | ||
| - [zuul/zuul] 859280: Trace Gerrit connection events https://review.opendev.org/c/zuul/zuul/+/859280 | ||
| -@gerrit:opendev.org- Zuul merged on behalf of Simon Westphahl: | 06:49 | |
| - [zuul/zuul] 859281: Trace Gitlab connection events https://review.opendev.org/c/zuul/zuul/+/859281 | ||
| - [zuul/zuul] 859282: Trace Pagure connection events https://review.opendev.org/c/zuul/zuul/+/859282 | ||
| -@gerrit:opendev.org- Dr. Jens Harbott proposed: [zuul/nodepool] 860213: DNM: Test quota failure on static driver https://review.opendev.org/c/zuul/nodepool/+/860213 | 08:11 | |
| -@gerrit:opendev.org- Dr. Jens Harbott proposed: [zuul/nodepool] 860213: DNM: Test quota failure on static driver https://review.opendev.org/c/zuul/nodepool/+/860213 | 08:32 | |
| @q:fricklercloud.de | I added a second static node to a pool, but it is not getting used, there is a pending node request which is always failing with "Not enough quota remaining to satisfy request", which shouldn't happen because the static driver cannot actually launch nodes. Any idea how to best debug this situation? | 08:41 |
|---|---|---|
| -@gerrit:opendev.org- Dr. Jens Harbott proposed: [zuul/nodepool] 860213: DNM: Test quota failure on static driver https://review.opendev.org/c/zuul/nodepool/+/860213 | 08:52 | |
| -@gerrit:opendev.org- Ian Wienand proposed: [zuul/zuul] 804956: web: Jobs: Use TreeView for job overview page https://review.opendev.org/c/zuul/zuul/+/804956 | 09:13 | |
| -@gerrit:opendev.org- Zuul merged on behalf of Simon Westphahl: [zuul/zuul] 859283: Trace Git connection events https://review.opendev.org/c/zuul/zuul/+/859283 | 10:29 | |
| -@gerrit:opendev.org- Zuul merged on behalf of Simon Westphahl: [zuul/zuul] 859284: Trace timer trigger events https://review.opendev.org/c/zuul/zuul/+/859284 | 10:30 | |
| -@gerrit:opendev.org- Zuul merged on behalf of Simon Westphahl: [zuul/zuul] 859285: Trace Zuul trigger events https://review.opendev.org/c/zuul/zuul/+/859285 | 10:37 | |
| @noonedeadpunk:matrix.org | Hey there! Sorry, I returned back to playing with zuul 4.9.0 here... And back to the issue of how to set a variable for job that can be used in hosts for the job | 13:26 |
| @noonedeadpunk:matrix.org | Eventually, according to the docs, job.extra-vars should do the trick. | 13:26 |
| @noonedeadpunk:matrix.org | As https://review.opendev.org/c/zuul/zuul/+/546474/6/zuul/executor/server.py#1740 they're added to ansible as extra-args indeed | 13:27 |
| @noonedeadpunk:matrix.org | BUT, looking at current code they're not more | 13:27 |
| @noonedeadpunk:matrix.org | And that's what I see in runtime - I can't really define hosts to be as a variable | 13:27 |
| @noonedeadpunk:matrix.org | Ie https://opendev.org/zuul/zuul/src/tag/4.9.0/zuul/executor/server.py#L2393-L2395 | 13:31 |
| @noonedeadpunk:matrix.org | So this statement from the doc is simply not true anymore https://zuul-ci.org/docs/zuul/4.9.0/reference/job_def.html#attr-job.extra-vars | 13:32 |
| @noonedeadpunk:matrix.org | On master I can't even find where they are, except docs | 13:33 |
| @noonedeadpunk:matrix.org | And that is super confusing. How you ended up with such statement.... https://opendev.org/zuul/zuul/src/commit/013b9fbd5324459cd48e225cdd918846f91043f8/zuul/executor/server.py#L844-L845 | 13:34 |
| @noonedeadpunk:matrix.org | corvus: pinging you as seems you worked on that :) | 13:39 |
| @noonedeadpunk:matrix.org | Ah, probably I can run against localhost and make the first task as `block` which delegate where I need it.... | 13:48 |
| @noonedeadpunk:matrix.org | "neat" | 13:49 |
| @clarkb:matrix.org | Dmitriy Rabotyagov: so I think there are two things going on here. First you are trying to accomplish something specific that isn't clear to me. Can you try to elaborate on what your goal is and why you think you need to copy playbooks around? The second is there may be a bug in the docs or the executor handling of extra vars. Unfortunately, it is unlikely that many here will debug such an old version of zuul, but we can check if modern zuul has the same problem. | 13:53 |
| @noonedeadpunk:matrix.org | So, with current code job.extra-vars are exactly same as job.host-vars. So they don't make any sense as a standalone thing. But, they were introduced and designed to have prescedence over host_vars and basically were loaded before inventory. | 13:55 |
| @noonedeadpunk:matrix.org | So if you want to run smth like that in job | 13:56 |
| ``` | ||
| - hosts: "{{ deploy_host | default('all[0]') }}" | ||
| become: true | ||
| tasks: | ||
| - name: Update ca-certificates | ||
| package: | ||
| name: ca-certificates | ||
| state: latest | ||
| update_cache: true | ||
| tags: | ||
| - skip_ansible_lint | ||
| ``` | ||
| @noonedeadpunk:matrix.org | There's simply no way to define `deploy_host` so it was respected | 13:57 |
| @noonedeadpunk:matrix.org | Now, in order to have different targets for post-merge tasks I can't define variable in jobs, so the 2 ways is either to copy code or do smth like | 13:58 |
| ``` | ||
| - hosts: localhost | ||
| become: true | ||
| tasks: | ||
| - name: We need that block jsut to workaround Zuul imperfectness | ||
| delegate_to: "{{ deploy_host | default('all[0]') }}" | ||
| block: | ||
| - name: Update ca-certificates | ||
| package: | ||
| name: ca-certificates | ||
| state: latest | ||
| update_cache: true | ||
| tags: | ||
| - skip_ansible_lint | ||
| ``` | ||
| @clarkb:matrix.org | My first thought is you should be using a group for that. | 13:59 |
| @clarkb:matrix.org | Regardless of whether or not extra vars handling is buggy Ansible groups exist for this use case | 14:00 |
| @noonedeadpunk:matrix.org | ah, you mean static and abstract name of the group... And jsut add hosts there regardless... | 14:01 |
| @jim:acmegating.com | Dmitriy Rabotyagov: i have a company that provides support for zuul, i'd be happy to talk to you about upgrading and finding solutions to issues like this! :) just DM me. | 14:02 |
| @noonedeadpunk:matrix.org | Well, yeah, that can work. But am I right that each playbook launched independently, so I will need to create/populate that group for each playbook in run? | 14:02 |
| @noonedeadpunk:matrix.org | corvus:you did that deployment ;) | 14:03 |
| @noonedeadpunk:matrix.org | but yes, I think I need to raise deployment upgrade topic internally as well | 14:04 |
| @noonedeadpunk:matrix.org | Clark: thank for the idea with groups, I was not considering them from this angle | 14:05 |
| @noonedeadpunk:matrix.org | I kind of knew that these groups are only in-memory, so defining it as a separate step in pre-run was not an option, so I dropped this idea away. But it's way better then block alternative anyway | 14:06 |
| @clarkb:matrix.org | You can also define group membership as part of your nodeset | 14:11 |
| @jim:acmegating.com | Dmitriy Rabotyagov: the change to how extra-vars is handled is directly related to the security vulnerabilities that were fixed in later versions. | 14:20 |
| @fungicide:matrix.org | > <@clarkb:matrix.org> You can also define group membership as part of your nodeset | 14:22 |
| and that persists across playbooks/phases, through the entirety of the build | ||
| @noonedeadpunk:matrix.org | > <@clarkb:matrix.org> You can also define group membership as part of your nodeset | 14:22 |
| well, these hosts are not part of nodepool | ||
| @fungicide:matrix.org | is this some sort of continuous deployment scenario, where the hosts are persistent? in opendev, we end up tackling that entirely with nested ansible through a deployment bastion, which the executor connects to | 14:24 |
| @noonedeadpunk:matrix.org | Well, I was thinking to define them just as variables for project job, so just override vars for the job based on the repo | 14:26 |
| @noonedeadpunk:matrix.org | And hoping to use native ansible as well after upgrade of zuul | 14:27 |
| @noonedeadpunk:matrix.org | Eventually, managing these hosts as static nodepool nodes also an option, but I don't really want any job to be able to access them. And when they're standalone I can use project specific ssh key that's in ssh-agent anyway | 14:28 |
| @clarkb:matrix.org | add_host allows you to specify group membership too | 14:30 |
| @noonedeadpunk:matrix.org | Yeah, thanks, as I explained I dropped that idea some time ago because failed to see where it's useful :D Still I think at very least documentation should be updated. | 14:34 |
| @noonedeadpunk:matrix.org | I will try to push smth that explains | 14:34 |
| @clarkb:matrix.org | Right I think there may be a bug here. But I wanted to decouple that specific behavior from what you are trying to accomplish. I believe what you want to do is achievable using normal Ansible groups. But without more background I can't say for sure | 14:38 |
| @jim:acmegating.com | the extra vars handling is no more than exactly what it says in the docs -- it's a higher precedence normal variable. it does not get passed to ansible as "--extra-vars". | 14:40 |
| @jim:acmegating.com | see https://zuul-ci.org/docs/zuul/latest/releasenotes.html#relnotes-4-6-0-security-issues for the upgrade note | 14:41 |
| @jim:acmegating.com | * the extra vars handling is no more than exactly what it says in the docs -- it's a higher precedence normal variable. it does not get passed to ansible with "-e". | 14:42 |
| @clarkb:matrix.org | Should the extra vars docs be updated to indicate they don't use -e to avoid confusion? | 14:55 |
| @jim:acmegating.com | no objection from me | 14:55 |
| -@gerrit:opendev.org- Clark Boylan proposed: [zuul/zuul] 860280: Clarify extra vars are not passed with -e https://review.opendev.org/c/zuul/zuul/+/860280 | 16:39 | |
| @jim:acmegating.com | zuul-maint: how does this look for a zuul release? | 17:28 |
| commit a0ba200ba70ce137780db397d681b1c3bd071fb8 (tag: 7.0.0) | ||
| @jim:acmegating.com | (i believe that's what opendev is running now) | 17:28 |
| @clarkb:matrix.org | Looking at the etherpad again I guess it is just 7.1.0 that is waiting on the openstack release (so that opendev can bump to ansible 6 everywhere and give that a good testing) | 17:34 |
| @jim:acmegating.com | Clark: yeah, and i think zuul has slowed down enough i don't think scheduling is going to be a problem (can you paste the link? i just restarted my browser and it will save me some digging) | 17:36 |
| @clarkb:matrix.org | https://etherpad.opendev.org/p/qF7VE9HzqPVzsCyZLWxb | 17:36 |
| @clarkb:matrix.org | I'm just trying to double check that sha now but otehrwise the plan sounds good to me | 17:37 |
| @clarkb:matrix.org | ok confirmed https://review.opendev.org/c/zuul/zuul/+/857065 was the last chagne to land before the weekend opendev updates and that merge sha maps to this change | 17:39 |
| -@gerrit:opendev.org- James E. Blair https://matrix.to/#/@jim:acmegating.com proposed: | 18:21 | |
| - [zuul/zuul-jobs] 860293: WIP: Add ensure-pip-localhost job https://review.opendev.org/c/zuul/zuul-jobs/+/860293 | ||
| - [zuul/zuul-jobs] 860294: More narrowly tailor the ensure-pip Debian workaround https://review.opendev.org/c/zuul/zuul-jobs/+/860294 | ||
| -@gerrit:opendev.org- James E. Blair https://matrix.to/#/@jim:acmegating.com proposed: [zuul/zuul-jobs] 860294: More narrowly tailor the ensure-pip Debian workaround https://review.opendev.org/c/zuul/zuul-jobs/+/860294 | 18:36 | |
| @jim:acmegating.com | Clark: ^ that's those two changes squashed (i wanted to confirm a failing test run before the change; looks like it works as expected. | 18:36 |
| @jim:acmegating.com | * Clark: ^ that's those two changes squashed (i wanted to confirm a failing test run before the change; looks like it behaves as expected. | 18:36 |
| @clarkb:matrix.org | ok I'll review the combo now | 18:37 |
| @jim:acmegating.com | thx! | 18:37 |
| @clarkb:matrix.org | corvus: is the shell task to cd into the zuul client git repo dir valid on localhost? | 18:39 |
| @jim:acmegating.com | Clark: yes, ansible_user_dir is set to the workspace dir | 18:40 |
| @jim:acmegating.com | (ie, $HOME is /var/lib/zuul/build/uuid/work/) | 18:41 |
| @clarkb:matrix.org | interesting | 18:41 |
| @clarkb:matrix.org | that was my only concern. I've approved it | 18:41 |
| @jim:acmegating.com | thanks! | 18:41 |
| -@gerrit:opendev.org- James E. Blair https://matrix.to/#/@jim:acmegating.com proposed: [zuul/zuul-jobs] 860296: Combine ensure-pip playbooks into a test role https://review.opendev.org/c/zuul/zuul-jobs/+/860296 | 18:47 | |
| @jim:acmegating.com | that should make maintenance easier, at the expense of some slightly longer paths | 18:47 |
| -@gerrit:opendev.org- Zuul merged on behalf of James E. Blair https://matrix.to/#/@jim:acmegating.com: [zuul/zuul-jobs] 860294: More narrowly tailor the ensure-pip Debian workaround https://review.opendev.org/c/zuul/zuul-jobs/+/860294 | 18:52 | |
| @vlotorev:matrix.org | > <@jim:acmegating.com> zuul-maint: how does this look for a zuul release? | 22:09 |
| > commit a0ba200ba70ce137780db397d681b1c3bd071fb8 (tag: 7.0.0) | ||
| > | ||
| Hi, any chance for nodepool release 7.0.0 sync? Last nodepool release was 6.2.0. (nodepool has a couple fixes for static nodes driver). | ||
| @jim:acmegating.com | we'll need a release note to explain the version bump (like we did for 5) | 22:52 |
| @jim:acmegating.com | feel free to propose that, and we can cut the release when it's there. or we could wait a week and just bump nodepool to 8.0.0 | 22:54 |
| @jim:acmegating.com | would likely save doing it twice :) | 22:54 |
| @jim:acmegating.com | zuul v7 release email sent | 22:55 |
| @vlotorev:matrix.org | It seems to me that previously nodepool version was bumped "just to be in sync with zuul version so that it would be easier for users to recognize which zuul/nodepool versions to upgrade", but I might be wrong about that. | 23:27 |
| I was not aware about nodepool 8.0 coming in a week :) | ||
| @clarkb:matrix.org | > <@vlotorev:matrix.org> It seems to me that previously nodepool version was bumped "just to be in sync with zuul version so that it would be easier for users to recognize which zuul/nodepool versions to upgrade", but I might be wrong about that. | 23:51 |
| > | ||
| > I was not aware about nodepool 8.0 coming in a week :) | ||
| The plan is to do zuul 8.0.0 when we drop support for Ansible 5. I think eh suggestion was to sync up nodepool at that point to avoid doing two sync up release notes | ||
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!