| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2020-07-09 | |||
| 08:32:13 | gibi | s/of/off/ | |
| 08:32:34 | gibi | I guess others from EU will do similar things | |
| 08:32:40 | brinzhang0 | gibi: have a holiday ^ | |
| 08:32:49 | gibi | thanks :) | |
| 08:34:56 | lyarwood | elod: https://review.opendev.org/#/c/739608/ should finally fix stable/rocky | |
| 08:44:04 | stephenfin | gibi, elod, lyarwood: Can raise on the team meeting, but what do we think about adding the Backport-Candidate flag to Gerrit like oslo has done? https://review.opendev.org/#/c/740068/ | |
| 08:46:44 | lyarwood | stephenfin: OOoOOOoO shiny | |
| 08:46:58 | lyarwood | stephenfin: +1 from me if we can query on that flag | |
| 08:47:15 | lyarwood | that we must be able to | |
| 08:47:33 | stephenfin | https://review.opendev.org/#/q/label:Backport-Candidate%253D%252B2 | |
| 08:47:34 | stephenfin | Yup | |
| 08:47:39 | stephenfin | (label:Backport-Candidate=+2) | |
| 08:48:37 | gibi | stephenfin: how does this relate to the way we target bugs to series in launchpad? | |
| 08:48:50 | gibi | will we do duble accounting? | |
| 08:48:53 | stephenfin | gibi: what do you mean? | |
| 08:49:24 | stephenfin | oh, as in setting the affected series for a reported bug in LP? | |
| 08:49:36 | gibi | in launchpad I can go to a bug and target it to multpile series (Ussurit, Train...) this way showing that the bugfix needs to be backported to those releases | |
| 08:50:04 | gibi | this also shows _where_ to backport | |
| 08:50:04 | stephenfin | that would sound like double accounting, yes. I guess we could either check both, or decide on which one we'd prefer | |
| 08:50:50 | stephenfin | iiuc we only do that launchpad tracking for some bugs, so we could use the backport-candidate feature for everything, and then be more specific in launchpad if necessary | |
| 08:53:33 | gibi | correct, there are non-bugs that we backport | |
| 08:53:50 | openstackgerrit | Lee Yarwood proposed openstack/nova stable/queens: Fix os_CODENAME detection and repo refresh during ceph tests https://review.opendev.org/740193 | |
| 08:53:52 | gibi | and launchpad does not help there | |
| 08:54:38 | gibi | do you have a good working example for a patch that is backported but would be strange to track it with an LP bug? | |
| 08:55:37 | stephenfin | I wasn't even being that specific. I meant bug reporters often don't set target versions in Launchpad, and I don't tend to set them when reviewing bugfixes | |
| 08:56:30 | stephenfin | But the ability to say "this looks like a fix we probably want to backport" from Gerrit does sound like something I'd use, if only because it doesn't involve context switching | |
| 08:57:34 | stephenfin | so tl;dr: this feels like something we'd collectively use far more, if that makes sense? | |
| 09:01:10 | gibi | stephenfin: I see, so the benefit of the new solution is that the recording of the information is closer to the place where the information is created. That definitely helps | |
| 09:01:42 | stephenfin | Yeah, that's a better way of saying what I'm trying to say :) | |
| 09:02:01 | xiaolin | lyarwood: Do you have time to review https://review.opendev.org/#/c/740151/ , I don not know how to change the branch and proposed a new one against branch master | |
| 09:02:11 | gibi | we can use the gerrit flag, to force a discusson of the question "how far to be backported?" and then the answer can be recorded in LP by assigning series | |
| 09:02:55 | lyarwood | xiaolin: I'll take a look today yes | |
| 09:03:33 | gibi | I'm not against the new flag, I feel it has some cost of re-learning the gerrit interface due to the extra flag a the reply button, but I see the benefit of the flag | |
| 09:03:44 | xiaolin | lyarwood: thanks | |
| 09:10:11 | stephenfin | gibi: I'm already doing that for oslo so not a massive deal for me, personally :) | |
| 09:10:50 | gibi | yeah, I can adapt too :) | |
| 09:31:07 | elod | stephenfin: that could help, yes, it's easy to see if a patch should be backported, in that way | |
| 09:31:44 | elod | stephenfin: i usually check the bugreport as well if a patch was marked to be backported to certain branches, too | |
| 09:32:21 | elod | stephenfin: unfortunately most of the time that detail is not filled in | |
| 09:34:53 | elod | stephenfin: so maybe it's worth adding that option :) | |
| 09:38:40 | hrw | morning | |
| 09:38:57 | hrw | https://1d9789842ee1597bfc07-adfe1c2e7945e3f1ebfe34ebeb2d7062.ssl.cf2.rackcdn.com/739800/2/check/kayobe-overcloud-centos8/c7c8c01/primary/kolla/nova/nova-compute.txt shows new nova issue | |
| 09:39:03 | hrw | 2020-07-08 16:44:28.078 6 ERROR nova File "/usr/lib/python3.6/site-packages/nova/virt/libvirt/utils.py", line 65, in <module> | |
| 09:39:06 | hrw | 2020-07-08 16:44:28.078 6 ERROR nova 'avx512vbmi': os_traits.HW_CPU_X86_AVX512VBMI, | |
| 09:39:09 | hrw | 2020-07-08 16:44:28.078 6 ERROR nova AttributeError: module 'os_traits' has no attribute 'HW_CPU_X86_AVX512VBMI' | |
| 09:43:57 | frickler | hrw: guess you either need an intermediate os_traits release or use it from git instead pypi | |
| 09:45:48 | hrw | or update to 2.4.0 | |
| 09:48:47 | lyarwood | yeah FWIW tripleo hit this as well and they just needed to pull in the new os-traits release | |
| 09:48:55 | lyarwood | that's AFAIK anyway | |
| 09:49:35 | hrw | weird as image says 2.4.0 os_traits. | |
| 09:49:41 | hrw | something to debug then | |
| 09:52:36 | bauzas | stephenfin: jumping late on your Backport-Candidate question, could you please explain the usage ? | |
| 09:53:10 | lyarwood | hrw: do you have a link to the build logs showing that? | |
| 09:53:53 | bauzas | stephenfin: if it's for telling "I think it's a good change for backporting it", then anyone can just bakcport it and leave the stable cores to look at it | |
| 09:54:08 | bauzas | I don't really see what it's helping | |
| 09:54:38 | hrw | lyarwood: https://1d9789842ee1597bfc07-adfe1c2e7945e3f1ebfe34ebeb2d7062.ssl.cf2.rackcdn.com/739800/2/check/kayobe-overcloud-centos8/c7c8c01/ is zuul job log dir | |
| 09:55:04 | hrw | lyarwood: docker kolla/centos-binary-nova-compute:master image used | |
| 09:55:05 | bauzas | stephenfin: and like gibi said, we already have both launchpad tags and series for stable | |
| 09:55:41 | hrw | yaawang: I did docker run image, python3, import os_traits and HW_CPU_X86_AVX512VBMI was there so wondering what is going on | |
| 09:56:13 | stephenfin | bauzas: You can just do the backport, but that means you have to remember to do it there and then | |
| 09:56:32 | stephenfin | which might be easier said than done, particularly if there are merge conflicts | |
| 09:57:00 | stephenfin | this is just a way for (core) reviewers to say "I think this should be backported" | |
| 09:57:16 | stephenfin | or, conversely, I think this should not be backported | |
| 09:57:27 | stephenfin | and then the actual backport can take place later | |
| 09:57:56 | hrw | stephenfin: 'Backport Candidate' options in gerrit you discuss? | |
| 09:58:03 | stephenfin | hrw: correct | |
| 09:58:17 | hrw | I like them. Use them in kolla | |
| 09:59:10 | stephenfin | So do I. Maybe you can answer the questions gibi and bauzas have? Namely, what it gives over setting the series metadata in Launchpad | |
| 10:00:54 | bauzas | mmm, I see your usecase then | |
| 10:01:03 | bauzas | it would be for people wanting to know what to backport ? | |
| 10:01:11 | bauzas | but honestly, who ? | |
| 10:01:19 | hrw | stephenfin bauzas: when I have fix for stable/* I have to get it into master and then u->t->s->r if needed. so I send for master and then comment with "backport needed" and say which branches | |
| 10:01:33 | hrw | so later it is visible which patches need backporting | |
| 10:01:38 | openstackgerrit | XiaolinZhao proposed openstack/nova master: Add USB keyboard to support for MIPS architecture https://review.opendev.org/740213 | |
| 10:01:56 | bauzas | and who would say +1/-1 ? stable cores or master cores ? | |
| 10:02:16 | hrw | bauzas: in kolla we have just cores | |
| 10:02:17 | bauzas | either way, not saying I disagree, just it looks to me something we don't really *need* | |
| 10:02:21 | stephenfin | bauzas: I mean, if I was given a list of stuff to backport and promised reviews then I'd be happy to skim the backport candidate list occasionally | |
| 10:02:42 | stephenfin | and it's backport candidate - not actual backport - so normal cores | |
| 10:02:48 | stephenfin | though we could conceivably do both | |
| 10:03:13 | hrw | bauzas: as you have master and stable cores then I would say that master core suggests backport and stable core may accept/reject | |
| 10:03:24 | stephenfin | yeah, that ^ | |
| 10:03:37 | bauzas | okay I see, it's then just yet another process... | |
| 10:04:01 | bauzas | some people would use it, some not | |
| 10:04:07 | stephenfin | Yes and no. It's just another checkbox in the review dropdown | |
| 10:04:24 | stephenfin | Personally, everyone is using it in oslo land because it's really low friction | |
| 10:04:43 | stephenfin | Certainly less than going to Launchpad and setting metadata there (which I don't currently do because I didn't know I was supposed to) | |
| 10:05:02 | hrw | xiaolin: nice to see that some other arch follows aarch64 steps | |
| 10:06:15 | stephenfin | bauzas: If it means we remember to backport even a few more bugfixes then I think it's a good thing, personally | |
| 10:06:22 | stephenfin | lyarwood: Have you been looking at https://review.opendev.org/739349 ? | |
| 10:06:37 | bauzas | okay okay, let's discuss tonight anyway | |
| 10:06:46 | stephenfin | seems tangentially related to stuff you've been working on recently | |
| 10:09:21 | lyarwood | stephenfin: on my list | |
| 10:10:15 | xiaolin | hrw: thanks : ) We just started | |
| 10:11:51 | hrw | xiaolin: check https://review.opendev.org/#/c/538003/ as you need to add test cases | |
| 10:17:41 | xiaolin | hrw: Thanks :) I will add test cases | |
| 10:21:00 | hrw | xiaolin: it took some revisions before I got them right so prefer to point you at start | |
| 10:48:37 | xiaolin | hrw: Thanks so much | |
| 10:50:41 | xiaolin | hrw: do you have time to review https://review.opendev.org/#/c/740151/ | |
| 10:52:10 | hrw | sure | |