| TheJulia | I'd appreciate any reviews on https://review.opendev.org/c/openstack/diskimage-builder/+/1009353 so I can unbrick ironic-python-agent-builder ci | 13:04 |
|---|---|---|
| clarkb | TheJulia: stevebaker[m] the problem with that approach is it will break anything that was relying on yum-utils | 15:51 |
| clarkb | ideally they would also declare that dependency but I think the safest approach is the one I described yesterday | 15:51 |
| TheJulia | clarkb: if you look at steve's latest patch before the stable branch, he introduced it | 16:17 |
| TheJulia | so, recent change created the break itself | 16:18 |
| TheJulia | oddly enough, I guess we don't invoke the epel element in dib runs itself | 16:18 |
| clarkb | TheJulia: yes but yum-utils was always insalled before | 16:19 |
| clarkb | thsi is a regression to stop installing it | 16:19 |
| clarkb | I understand that may fix your use case but it potentially breaks others | 16:19 |
| clarkb | the solution is to only install it on platforms that actually use epel and I outlined how to do that in a way that doesn't require extra bookkeeping | 16:20 |
| clarkb | actually the old code used this approach | 16:21 |
| clarkb | so I think maybe we should just revert | 16:21 |
| clarkb | or do a partial revert. I think that change did some other stuff too | 16:21 |
| TheJulia | Yeah, basically | 16:21 |
| TheJulia | I've got a freight delivery showing up any minute, once my shiny new batteries are secured, I'll rev the patch and take a partial revert path. | 16:34 |
| clarkb | ack I'll be around and should be able to review it quickly | 16:38 |
| TheJulia | k, thx | 16:54 |
| opendevreview | Julia Kreger proposed openstack/diskimage-builder master: epel: partial revert of Ic6f6c92b28b80c4f0b93e4663129e73134f5b447 https://review.opendev.org/c/openstack/diskimage-builder/+/1009353 | 17:33 |
| TheJulia | I think that does it | 17:33 |
| clarkb | TheJulia: I +2'd the noticed that its DIB_EPEL_DISABLED==2 where we skip the work | 17:39 |
| clarkb | so I think both 0 and 1 need to install yum-utils? | 17:39 |
| clarkb | so the condition should be changed to DIB_EPEL_DISABLED != 2 ? | 17:39 |
| clarkb | anyway posted a comment about that too | 17:39 |
| TheJulia | I guess I'm not seeing why != 2 when the other logic is intended to bomb when it is != 0 | 17:40 |
| clarkb | when it is 0 we enable epel fully. When it is 1 we install epel but don't enable it by default. when it is 2 we don't do anything | 17:40 |
| clarkb | (really you should just drop the element in that case but I'm trying to work with users here) | 17:40 |
| TheJulia | okay | 17:41 |
| TheJulia | I misread the prior change then | 17:41 |
| TheJulia | I was looking at line 3's gt 0 and not line 9's gt 1 | 17:41 |
| TheJulia | fixing | 17:41 |
| opendevreview | Julia Kreger proposed openstack/diskimage-builder master: epel: partial revert of Ic6f6c92b28b80c4f0b93e4663129e73134f5b447 https://review.opendev.org/c/openstack/diskimage-builder/+/1009353 | 17:41 |
| TheJulia | done | 17:41 |
| * TheJulia context switches back | 17:42 | |
| clarkb | +2 thanks | 17:42 |
| clarkb | if no one else weighs in while it is in check you can probably go ahead and approve it | 17:42 |
| TheJulia | k | 17:42 |
| stevebaker[m] | It looks like we need an environment.d script to set a default value for DIB_EPEL_DISABLED https://zuul.opendev.org/t/openstack/build/037dd6a5985b4e57a3f93a82ce82372b/log/job-output.txt#1446-1458 | 19:48 |
| opendevreview | Julia Kreger proposed openstack/diskimage-builder master: epel: partial revert of Ic6f6c92b28b80c4f0b93e4663129e73134f5b447 https://review.opendev.org/c/openstack/diskimage-builder/+/1009353 | 19:52 |
| TheJulia | Yeah, I was just doing that :) | 19:52 |
| TheJulia | I was just making sure the functests failed for the same reason | 19:52 |
| opendevreview | Julia Kreger proposed openstack/diskimage-builder master: epel: partial revert of Ic6f6c92b28b80c4f0b93e4663129e73134f5b447 https://review.opendev.org/c/openstack/diskimage-builder/+/1009353 | 19:53 |
| TheJulia | okay, now I really mean it | 19:53 |
| TheJulia | so, passed CI with that revision, seems to be solid. clarkb a quick re-review would be appreciated | 21:41 |
| TheJulia | fwiw, 3 jobs do fail, but it appears to be unrelated playbooks are expecting disk-image-create to exist which means the NV jobs are broken in entirely unrelated way | 21:42 |
| clarkb | on it | 21:42 |
| clarkb | ya suse for sure hasn't passed in a while and maybe we should drop it. JayF was keeping gentoo working for a while. The Debian one surprises me but is non voting and won't touch epel so its fine | 21:44 |
| JayF | the gentoo one should probably be moved to experimental | 21:44 |
| JayF | keeping a gentoo DIB CI working with gentoo being a rolling release is more effort than it's worth, but ensuring the CI passes once in a while is good as a "sync" back to latest | 21:45 |
| clarkb | makes sense | 21:45 |
| opendevreview | Jay Faulkner proposed openstack/diskimage-builder master: Move gentoo tests to experimental https://review.opendev.org/c/openstack/diskimage-builder/+/1009643 | 21:53 |
| opendevreview | Merged openstack/diskimage-builder master: epel: partial revert of Ic6f6c92b28b80c4f0b93e4663129e73134f5b447 https://review.opendev.org/c/openstack/diskimage-builder/+/1009353 | 22:32 |
Generated by irclog2html.py 4.1.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!