| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2017-10-26 | |||
| 13:44:27 | mriedem | stvnoyes: ok, alternatively i can just add something to a tempest test, push that up and make it depend on the nova api change | |
| 13:44:45 | stvnoyes | having it in tempest would be good to have | |
| 13:44:57 | mriedem | ok i'll put something up this morning | |
| 13:46:05 | openstackgerrit | Eric Fried proposed openstack/nova master: Parse granular resources/traits from extra_specs https://review.openstack.org/515151 | |
| 13:46:06 | openstackgerrit | Eric Fried proposed openstack/nova master: Granularize resources_from_{flavor|request_spec} https://review.openstack.org/515223 | |
| 13:46:06 | cdent | edleafe, efried, alex_xu: are you familiar with the “park on a slope” concept, when experiencing task interruptions? | |
| 13:46:07 | efried | cdent I went ahead and made those couple of changes since I was editing the base patch anyway ^ | |
| 13:46:26 | efried | cdent Not by that name. Googling... | |
| 13:46:44 | edleafe | cdent: sorry, you just interrupted my task | |
| 13:46:45 | cdent | efried: cool, given other commitments today it is unlikely I would have got back to it and/or there’s not enough coffee in the world to make my brain work | |
| 13:46:52 | edleafe | cdent: no, not familiar | |
| 13:47:31 | efried | Turns out that's kind of a tough thing to google. | |
| 13:48:28 | cdent | it’s aligned with this idea of leaving a TODO in the wip commit. park your car in a way that if the battery is flat when you get back to it, you can still rolling start it | |
| 13:48:44 | cdent | or in general: make it easy, by leaving info, to rolling start | |
| 13:48:48 | cdent | stigmergy | |
| 13:48:50 | openstackgerrit | Ghanshyam Mann proposed openstack/nova master: Implement query param schema for flavor index https://review.openstack.org/515413 | |
| 13:49:14 | cdent | efried: google stigmergy if you want something detailed to google | |
| 13:50:23 | efried | cdent Neat - the ants-following-scent-trails thing was a great way to crystallize the concept in my head. | |
| 13:50:53 | edleafe | efried: that's what we are, after all :) | |
| 13:51:15 | efried | cdent edleafe Not sure I would go so far as to say we "lack any memory, intelligence or even individual awareness of each other" though. | |
| 13:51:44 | edleafe | efried: so effin' literal! | |
| 13:52:09 | cdent | efried: that somehow reminded me of an old blog post: https://www.burningchrome.com/~cdent/mt/archives/000187.html | |
| 13:52:45 | cdent | efried: if we want to be an inclusive community that accepts all comers we should behave in a way that assumes there is no shared memory, that only external things can operate as memory | |
| 13:53:25 | efried | *mind blown* | |
| 13:55:41 | edleafe | efried: that's cdent's fancy way of saying "write it down!" | |
| 13:57:23 | bauzas | sahid: as you pointed out, there is a bug that prevents mediated devices to be shown by libvirt https://bugzilla.redhat.com/show_bug.cgi?id=1463285#c6 | |
| 13:57:24 | openstack | bugzilla.redhat.com bug 1463285 in libvirt "mediated devices are not shown in nodedev-list --cap mdev output" [High,Post] - Assigned to eskultet | |
| 13:57:50 | bauzas | sahid: now, the problem is about which specific operator impact we should provide | |
| 13:58:11 | stephenfin | dansmith: Wanna direct your nova-network wrath on this? https://review.openstack.org/#/c/499168/ | |
| 13:58:13 | bauzas | sahid: if we ask operators to pre-create the mdevs, it can work for the next instances to be created | |
| 13:58:54 | dansmith | stephenfin: I don't really have n-net wrath, but sure :) | |
| 13:58:58 | bauzas | sahid: but for existing instances, it can be an operator painpoint because you need to exactly recreate the mdev with the same existing UUID | |
| 13:59:05 | dansmith | anything to avoid working on this slide deck sitting in front of me | |
| 13:59:11 | bauzas | as reboots wipe out mediated devices | |
| 13:59:33 | bauzas | sahid: that's why I feel we should leave nova drive the creation, and not the operator | |
| 13:59:45 | bauzas | sahid: unless mediated devices become persistent upon reboot :) | |
| 14:00:08 | sahid | bauzas: what i could say is that, for SR-IOV we ask operator to allocate the VFs | |
| 14:00:16 | openstackgerrit | Alex Xu proposed openstack/nova master: Enable _get_usages_by_provider_and_rc not filter on the resource class id https://review.openstack.org/514649 | |
| 14:00:16 | openstackgerrit | Alex Xu proposed openstack/nova master: Add _get_traits_by_rp_ids method https://review.openstack.org/511184 | |
| 14:00:17 | openstackgerrit | Alex Xu proposed openstack/nova master: Add ProviderSummaryList object https://review.openstack.org/514198 | |
| 14:00:17 | openstackgerrit | Alex Xu proposed openstack/nova master: Add more caches for the ProviderSummary and ProviderSummaryList https://review.openstack.org/514199 | |
| 14:00:18 | openstackgerrit | Alex Xu proposed openstack/nova master: Add AggregatedProviderSummary and AggregatedProviderSummaryList objects https://review.openstack.org/514200 | |
| 14:00:18 | openstackgerrit | Alex Xu proposed openstack/nova master: Add has_resources method to the AggregatedProviderSummary object https://review.openstack.org/514650 | |
| 14:00:19 | openstackgerrit | Alex Xu proposed openstack/nova master: Add as_allocation_request_obj method to AggregatedProvierSummary obj https://review.openstack.org/514651 | |
| 14:00:19 | openstackgerrit | Alex Xu proposed openstack/nova master: placement: rewrite AllocationCandidates.get_by_filters https://review.openstack.org/479766 | |
| 14:00:20 | openstackgerrit | Eric Fried proposed openstack/nova master: Reformat _get_all_with_shared https://review.openstack.org/512799 | |
| 14:00:25 | sahid | if host is rebooting the operator can have the module to pre-allocate VFs anyway | |
| 14:00:37 | sahid | but without the insurance that the PCI addresses will be the same as previously | |
| 14:00:47 | bauzas | sahid: but the domain XML will be incorrect then, right? | |
| 14:00:51 | dansmith | stephenfin: can you fix that typo in the commit message and then I'll fast approve? | |
| 14:00:52 | sahid | right | |
| 14:01:00 | bauzas | sahid: if the VF is on the wrong PCI device | |
| 14:01:01 | sahid | so operator have to deal with that | |
| 14:01:09 | alex_xu | efried: ^ more todo, some of them is more interest than unittest, | |
| 14:01:09 | bauzas | sahid: how can they deal with that ? | |
| 14:01:21 | openstackgerrit | Stephen Finucane proposed openstack/nova master: conf: Move additional nova-net opts to 'network' https://review.openstack.org/499168 | |
| 14:01:22 | dansmith | stephenfin: or I can just do it in the ui I guess | |
| 14:01:22 | sahid | reboot --force | |
| 14:01:29 | stephenfin | dansmith: Done | |
| 14:01:30 | bauzas | sahid: they dump all the instances XML before starting nova-compute? | |
| 14:01:32 | stephenfin | ...from the UI :) | |
| 14:01:52 | efried | alex_xu Ack | |
| 14:01:54 | dansmith | stephenfin: done | |
| 14:01:59 | bauzas | sahid: but a reboot breaks existing instances using VFs, right? | |
| 14:02:29 | bauzas | since as you said, there is no guarantee that the newly created VF will match the one descibed in the domain XML | |
| 14:02:44 | sahid | bauzas: there are some limitations, it's clear... what i'm trying to say is, if you can make them clear for your implementation so operator could deal with them | |
| 14:03:55 | sahid | the libvirt design merged upstream for pci devices does not manage the devices | |
| 14:04:13 | bauzas | sahid: I'm thinking about the fact that if I'm implementing something like nova creating the mdevs directly, we could just ask the operator to restart their libvirt deamon right after they start nova-computez | |
| 14:04:14 | sahid | you could try to do it in nova | |
| 14:04:19 | sahid | but that seems to be complicated | |
| 14:04:48 | sahid | bauzas: what about to ask operator to do a force reboot? | |
| 14:05:01 | dansmith | bauzas: restarting libvirt after nova-compute isn't reasonable, IMHO | |
| 14:05:09 | dansmith | or starting | |
| 14:05:27 | sahid | we have to double check but with a force-reboot we recreating the domainXML from scratch | |
| 14:06:35 | alex_xu | cdent: yea, always be interrupted, like meeting... | |
| 14:06:35 | sahid | dansmith is right | |
| 14:06:53 | bauzas | dansmith: that's a separate problem | |
| 14:07:06 | bauzas | dansmith: there are two issues with libvirt managing mdevs | |
| 14:07:40 | bauzas | dansmith: #1 is that mdevs aren't persistent upon reboots | |
| 14:08:03 | bauzas | dansmith: #2 is that there is a kernel bug that prevents newly created mdevs to be seen by libvirt | |
| 14:08:39 | bauzas | for #1, my take is that we shouldn't leave the operators deal with the mess of recreating specific mdevs based on specific UUIDs | |
| 14:08:52 | bauzas | rather, it should be nova that should deal with that | |
| 14:09:06 | dansmith | yeah | |
| 14:09:08 | bauzas | for #2, it's just a kernel bug that is tackled by https://bugzilla.redhat.com/show_bug.cgi?id=1463285#c6 | |
| 14:09:09 | openstack | bugzilla.redhat.com bug 1463285 in libvirt "mediated devices are not shown in nodedev-list --cap mdev output" [High,Post] - Assigned to eskultet | |
| 14:09:23 | bauzas | then, it's unrelated to nova | |
| 14:09:35 | dansmith | if restarting libvirt is a requirement as a workaround for a bug that will be fixed, then that's fine | |
| 14:09:47 | bauzas | but we could workaround that by asking operators to restart their libvirt deamon | |
| 14:09:47 | openstackgerrit | Hongbin Lu proposed openstack/nova stable/pike: Handle exception on adding secgroup https://review.openstack.org/515418 | |
| 14:10:00 | dansmith | but yeah, if nova is creating mdevs with uuids and then we lose those on reboot, we have to recreate them, IMHO | |
| 14:10:03 | bauzas | dansmith: that's my point, it's just a stupid workaround | |
| 14:10:09 | dansmith | ack, makes sense | |
| 14:11:01 | bauzas | and with some hope, distributions would ship the right fixed kernel | |
| 14:11:28 | bauzas | or the fix could be backported in the supported kernel, either way | |
| 14:12:12 | bauzas | sahid: dansmith: wait, I have another idea | |
| 14:12:38 | bauzas | oVirt does get the mdev list by not calling the libvirt API, but rather directly checking the FS | |
| 14:12:58 | bauzas | that's why they're not impacted by the bug | |
| 14:13:14 | bauzas | that could be another idea | |
| 14:13:28 | bauzas | although I definitely prefer to use the libvirt API | |
| 14:13:40 | openstackgerrit | Hongbin Lu proposed openstack/nova stable/ocata: Handle exception on adding secgroup https://review.openstack.org/515421 | |