15:30:20 <jbair> #startmeeting Fedora CI SIG 15:30:20 <zodbot> Meeting started Wed Jan 27 15:30:20 2021 UTC. 15:30:20 <zodbot> This meeting is logged and archived in a public location. 15:30:20 <zodbot> The chair is jbair. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:30:20 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic. 15:30:20 <zodbot> The meeting name has been set to 'fedora_ci_sig' 15:30:20 <jbair> .hello jimbair 15:30:21 <zodbot> jbair: jimbair 'Jim Bair' <jbair@redhat.com> 15:33:19 <msrb> .hello msrb 15:33:20 <zodbot> msrb: msrb 'Michal Srb' <msrb@redhat.com> 15:34:42 <jbair> I do wonder if we need to move this to a different time; possibly we're conflicting with something :) 15:34:46 <jbair> We used to have far more attendance 15:34:58 <bgoncalv> .hello2 15:35:01 <zodbot> bgoncalv: bgoncalv 'Bruno Goncalves' <bgoncalv@redhat.com> 15:35:42 <bgoncalv> any specific topic to discuss? Maybe people just forgot about the meeting, we can try to ping them :) 15:36:09 <jbair> No specific topics today 15:36:15 <jbair> I am wondering if we need to find a better time for it? 15:36:24 <jbair> Maybe another Fedora meeting is conflicting 15:36:27 <jbair> like FESCo or something 15:37:22 <jbair> I know that's this morning (my time) but not sure exactly when 15:39:26 <jbair> I'll send a note after we wrap here about it to ci@ 15:39:44 <jbair> any topics you two wanted to discuss? I know we're still working on the new jenkins; are both 1 and 2 working? Or are we mostly only configured on 1? 15:40:33 <bgoncalv> I haven't touched Fedora CI for a while now, but yes I assume pipelines are working 15:40:49 <bgoncalv> pipeliens are running on jenkins-1, monitor bots on jenkins-2 15:41:23 <jbair> I haven't heard much, which is usually good news :) 15:41:51 <bgoncalv> jbair: since last meeting only infra issue are posted to this channel 15:42:22 <jbair> I didn't make any changes so I would guess someone else did? 15:42:34 <bgoncalv> I did, this was just FYI :) 15:42:41 <jbair> ah, well thanks :) 15:42:50 <jbair> It seems far less noisy, and I appreciate it 15:43:04 <msrb> yep, I asked Bruno to silence the test failure notifications ;) 15:43:13 <jbair> msrb: ty :) 15:47:00 <msrb> one thing worth sharing: 15:48:25 <jbair> what's that? 15:48:31 <msrb> people asked about gating on scratch-builds of dependent components; like pass gating if scratch-builds of dependent components were successful 15:48:44 <msrb> sounds like Koschei 15:49:18 <msrb> however, they only wanted a specific list of components (kernel right now, when glibc changes) 15:49:37 <jbair> so like kernel gating against scratch builds of glibc? 15:49:47 <msrb> so I tried to prototype it today: https://bodhi.fedoraproject.org/updates/FEDORA-2021-6c71425dab 15:50:17 <msrb> jbair, the other way around, glibc gating against kernel scratch-builds 15:51:14 <bgoncalv> I think greenwave doesn't support results for scratch builds... 15:51:29 <bgoncalv> even if resultsdb gets update to save it 15:51:37 <msrb> bgoncalv, kernel scratch build is reported as glibc test result 15:51:47 <bgoncalv> ahhh 15:51:55 <msrb> bgoncalv, with "scenario: kernel" 15:52:12 <msrb> but scenarios doesn't seem to be supported in resultsdb-updater yet 15:53:01 <msrb> so people should be able to add "!PassingTestCaseRule {test_case_name: fedora-ci.koji-build.build.static-analysys, scenario: kernel} to their gating.yaml 15:53:14 <msrb> and of course not only kernel, but any other component name as well 15:53:31 <msrb> but yeah, it doesn't work properly end-to-end yet 15:53:42 <bgoncalv> seems some sort of reverse dependency, but the revdep pipeline would create this scratch build 15:53:49 <msrb> yep 15:53:50 <msrb> exactly 15:53:54 <msrb> same thing basically 15:54:11 <bgoncalv> the new kernel build should use the new glibc 15:54:32 <bgoncalv> but the glibc will not be in buildroot, as didn't pass gating, right? 15:54:46 <msrb> yeah, I add new glibc to a side-tag and then scratch-build kernel in it 15:55:02 <msrb> that's what I am doing right now 15:55:50 <msrb> one big missing piece is configuration: how will people tell us which components to scratch-build 15:56:03 <bgoncalv> what about instead of using scenario, we use different testcase name? 15:56:07 <msrb> but we will need to figure out the same for regular revdeps pipeline as well 15:56:28 <bgoncalv> test_case_name: fedora-ci.koji-build.build.revdep-glibc-kernel ? 15:57:07 <bgoncalv> we could use same syntax used for explicit define revdep tests as used internally 15:57:15 <bgoncalv> on ci.fmf 15:57:55 <msrb> bgoncalv, yeah, unique test case name would work as well 15:58:14 <bgoncalv> should help us to waive and rerun only what we want 15:58:14 <msrb> I just thought that scenarios are a bit cleaner solution 15:58:50 <msrb> scenario is basically additional context for the test result 15:59:17 <bgoncalv> not sure how this work in the infra 15:59:18 <msrb> so "the same scratch-build test, but with package xyz" 15:59:27 <bgoncalv> like greenwave, bodhi interface... :) 15:59:39 <msrb> bgoncalv, bodhi is already able to visualize it 15:59:47 <bgoncalv> cool 16:01:15 <msrb> like here: https://bodhi.fedoraproject.org/updates/FEDORA-2021-b71b8a85c8 16:01:39 <msrb> those "x86_64 workstation" strings (should be) scenarios 16:01:57 <msrb> they *should* be waivable independently 16:02:13 <jbair> I can put this in my post-meeting email for discussion to see if people have any thoughts about it (a thing I plan on doing to try and boost attendance) 16:02:19 <msrb> and since all those are separate messages, people will be able to click and rerun only what they want 16:02:39 <msrb> jbair, go for it ;) 16:03:28 <bgoncalv> okay, nothing more to add :) 16:03:28 <jbair> I hopped into a 10AM meeting so I'll send it after that; give it a read and reply if I get any details wrong :) 16:03:54 <msrb> ;) 16:26:30 <jbair> okay thanks guys :) 16:26:36 <jbair> #endmeeting