| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2018-09-06 | |||
| 13:31:12 | naichuans | :) bauzas: | |
| 13:32:41 | openstack | Launchpad bug 1790847 in OpenStack Compute (nova) "Failing tests when building Rocky Debian package in Python 3.7" [Medium,Confirmed] | |
| 13:32:41 | mriedem | if anyone has a py37 system, we have a few unit test bugs https://bugs.launchpad.net/nova/+bug/1790847 https://bugs.launchpad.net/nova/+bug/1790849 | |
| 13:32:42 | openstack | Launchpad bug 1790849 in OpenStack Compute (nova) "Failng hacking tests when building Rocky Debian package under Python 3.7" [Medium,Confirmed] | |
| 13:33:23 | openstack | Launchpad bug 1771506 in OpenStack Compute (nova) "Unit test failure with OpenSSL 1.1.1" [Low,Confirmed] | |
| 13:33:23 | bauzas | naichuans: mriedem: yup, remember this https://bugs.launchpad.net/nova/+bug/1771506 | |
| 13:33:51 | bauzas | we're like super dependent on openssl because of xen | |
| 13:33:59 | bauzas | something we should probably mock out | |
| 13:35:10 | mriedem | ah i'll mark the dupe | |
| 13:35:49 | naichuans | Yes, it should be same. Thanks, bauzas | |
| 13:36:17 | naichuans | I will look into it tomorrow | |
| 13:41:47 | openstackgerrit | Eric Fried proposed openstack/nova master: Add an example to add more pci devices in nova.conf https://review.openstack.org/592243 | |
| 13:42:31 | efried | mriedem: I would feel more comfortable approving https://review.openstack.org/#/c/592243/ if I had your eyes on it. | |
| 13:42:42 | efried | stephenfin: wanna have a look at that one last time -^ | |
| 13:42:50 | efried | (hopefully last :) | |
| 13:47:27 | mriedem | efried: pci anything in nova is my biggest weakness | |
| 13:47:50 | mriedem | i added moshe and sahid | |
| 13:48:14 | efried | mriedem: I already shepherded this patch to not touch the pci code. It was originally going to change the logic so you *could* specify [{...}, ...] | |
| 13:48:22 | efried | mriedem: But okay, thanks. | |
| 13:50:12 | melwitt | nova meeting in 10 min | |
| 13:50:18 | mriedem | this reminds me to look at that persons other patch https://review.openstack.org/#/c/582899/ | |
| 13:50:25 | mriedem | with 0 explanation in the commit mesage | |
| 13:50:28 | mriedem | *message | |
| 13:55:22 | mriedem | big ol 01 | |
| 13:55:23 | mriedem | -1 | |
| 13:55:50 | mriedem | dansmith: heh i was just going to point you at ^ | |
| 13:56:01 | mriedem | and now see, "We should get Matt's read on this as well. However, I'm skeptical of how/why this would be happening and thus think we should require a healthy bug report showing the root cause. " | |
| 13:56:08 | mriedem | i said basically the same thing | |
| 13:56:30 | dansmith | is there more info now? | |
| 13:56:41 | mriedem | not in the commit message | |
| 13:57:05 | mriedem | the bug report says there are mq issues | |
| 13:57:08 | mriedem | that's about it | |
| 13:57:22 | dansmith | right, well, that's not enough, IMHO | |
| 13:57:36 | mriedem | yeah i can't tell if this is talking about resize or build reschedules or what | |
| 13:58:45 | mriedem | "Exception during message handling: MessageDeliveryFailure: Unable to connect to AMQP server on 10.127.3.64:5672 after None tries: 'NoneType' object has no attribute '__getitem__'" | |
| 13:59:12 | mriedem | doesn't really tell me anything | |
| 13:59:49 | dansmith | I'm just not interested in working too hard at it with so little detail | |
| 14:00:26 | mriedem | yeah me neither | |
| 14:00:37 | mriedem | i'm just not interested in working | |
| 14:03:08 | mriedem | damn gibi is gone | |
| 14:03:15 | mriedem | need a re-+2 on https://review.openstack.org/#/c/599744/ | |
| 14:03:41 | mriedem | alex_xu: or you can proxy gibi's +2 and approve ^ since he was +2 before i removed the try/except stuff | |
| 14:04:37 | openstackgerrit | Claudiu Belu proposed openstack/nova master: hyper-v: autospec classes before they are instantiated https://review.openstack.org/342211 | |
| 14:05:39 | dansmith | mriedem: so that online migration has to be removed when the split happens yeah? | |
| 14:07:20 | mriedem | dansmith: yup | |
| 14:07:30 | mriedem | i don't know how we're going to do those | |
| 14:07:31 | dansmith | ah, I guess this is all within placement tables anyway? | |
| 14:07:39 | mriedem | yeah | |
| 14:07:41 | dansmith | I was thinking this had to cross-reference nova and placement things | |
| 14:07:58 | dansmith | okay, so placement'll need a way to trigger this beyond that I suppose | |
| 14:08:35 | cdent | i've started some _very_ hand-wavey motions in the direction of a placement-manage | |
| 14:09:01 | mriedem | cdent: efried: the reshaper item ~L269 in https://etherpad.openstack.org/p/nova-ptg-stein should probably move to the retrospective etherpad? | |
| 14:09:06 | mriedem | it's more of a retrospective item right? | |
| 14:09:55 | efried | mriedem: okay | |
| 14:10:07 | sahid | efried: i shared my idea | |
| 14:10:10 | cdent | yeah, I think that existed before the retro etherpad did. and the retro etherpad needs friends | |
| 14:10:10 | efried | which, if we do that, will let cdent keep the reins | |
| 14:12:12 | mriedem | so i guess i'll move it | |
| 14:12:30 | mriedem | done | |
| 14:12:57 | efried | thanks matt | |
| 14:22:38 | openstackgerrit | Vladyslav Drok proposed openstack/nova master: Fix resource tracker updates during instance evacuation https://review.openstack.org/600425 | |
| 14:29:07 | stephenfin | efried: We shouldn't be dropping 'default' on that patch https://review.openstack.org/#/c/592243/ | |
| 14:38:32 | openstackgerrit | sean mooney proposed openstack/nova master: add caching to _build_regex_range https://review.openstack.org/599071 | |
| 14:42:11 | sean-k-mooney | efried: alex_xu ^ i have made this use just a dictonry as a cache. im still getting a reasonable speedup, we may want to consider going to the global cache at a later point but this is simpler solution for now | |
| 14:52:40 | mriedem | oh boy the age old "should every commit be deployable" question is being debated in the tc channel | |
| 14:53:50 | melwitt | recently penick was talking about having a new cluster that deploys trunk | |
| 15:00:19 | efried | stephenfin: The problem with the default being [] is that it makes you think you can use square brackets to define a list, which you can't. And back in PS2 where they were trying to fix the code so that it *would* accept square brackets, I told them to quit because we don't want to be spending more resources working on this code. Which of course we now are :( | |
| 15:01:00 | stephenfin | efried: Ah, this is the bug in the oslo.config sphinxext module, right? | |
| 15:01:10 | bauzas | efried: dansmith: mriedem: others, sorry had to leave urgently because I missed my kids at schoold | |
| 15:01:25 | bauzas | I'm fine with punting the question about PCI future at the pth | |
| 15:01:27 | bauzas | pth | |
| 15:01:29 | bauzas | shit | |
| 15:01:30 | bauzas | ptg | |
| 15:02:06 | bauzas | but my take on it is "we can do some generic device refactoring in nova, while cyborg raises up" | |
| 15:02:15 | bauzas | I don't think it's a problem to us | |
| 15:02:20 | bauzas | and the community | |
| 15:02:49 | efried | bauzas: Okay, please take a look at https://review.openstack.org/#/c/591037/ when you get a chance then. | |
| 15:03:18 | efried | stephenfin: I don't think it has to do with sphinxext | |
| 15:03:53 | efried | stephenfin: If the default is [] I would think the sphinx doc should display it as []. But that's the confusing part (to the user). | |
| 15:04:42 | mriedem | efried: btw, i don't mean to minimize your gpu/pci/cyborg thing, i was just airing out my anxiety over the amount of stuff we have loaded up | |
| 15:04:58 | mriedem | which mostly makes me want to crawl in a hole | |
| 15:07:19 | efried | mriedem: Duly noted. But I should point out that I am expected to be making certain things happen to the benefit of those who sign my paycheck. I.e. I'm not totally free to work on a thing just because it's deemed a priority for nova-at-large, if that is to the exclusion of working on a thing that's a slightly-lesser-priority-for-nova-but-#1-priority-for-employer. | |
| 15:07:56 | openstackgerrit | Matt Riedemann proposed openstack/nova stable/rocky: Configure placement DB context manager for nova-manage/status https://review.openstack.org/600464 | |
| 15:07:57 | mriedem | efried: yup i know, same here | |
| 15:08:53 | bauzas | heh, we're all on the same boat | |
| 15:09:05 | dansmith | heh | |
| 15:09:32 | bauzas | or you would see me more often | |
| 15:16:42 | mriedem | well, | |
| 15:16:48 | mriedem | we don't all get august off... | |
| 15:26:23 | bauzas | mriedem: touché, my point being on the 45 other weeks :p | |
| 15:30:29 | efried | stephenfin: Hm, actually it looks like there *is* a sphinx bug https://docs.openstack.org/nova/latest/configuration/config.html#pci <== the default is showing up as u'' here | |
| 15:31:08 | efried | I mean, it's still confusing if you're looking at the source; not sure how many admins do that vs looking at the html. | |
| 15:31:13 | efried | and if that bug gets fixed... | |
| 15:33:34 | dansmith | mriedem: looks like the latest rev of the configurable instance list behavior is failing unit tests | |
| 15:33:54 | dansmith | she's out until next week, but maybe we can get that landed then, if I don't get a chance to fix some trivial test failure before then | |
| 15:34:17 | mriedem | landed in denver you mean? | |
| 15:34:20 | mriedem | i saw the tests failing | |
| 15:35:22 | dansmith | landed in denver yeah | |
| 15:36:34 | dansmith | hmm, it's in the middle of her stack now, | |
| 15:36:38 | dansmith | which is probably why it's breaking | |