| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2019-02-26 | |||
| 10:31:51 | aspiers | kashyap: yeah, I noticed you were not in the channel | |
| 10:31:56 | aspiers | correct | |
| 10:32:09 | kashyap | aspiers: Which TZ are you in? | |
| 10:32:12 | aspiers | and this stack of patches: https://review.openstack.org/#/c/638680/ | |
| 10:32:18 | aspiers | GMT (London, UK) | |
| 10:33:34 | kashyap | Okido; will come back sometime in the afternoon to this. | |
| 10:33:46 | kashyap | aspiers: I'll leave you with a related URL: https://review.openstack.org/#/c/631154/ ("WIP: Gracefully handle QEMU machine types for guests") | |
| 10:36:42 | sean-k-mooney | stephenfin: looking at the other docs changes you pushed im more or less hapy with the final too in the series. ill take a look at them again when the ci reports back but do you want to include them in the release? | |
| 10:36:43 | aspiers | kashyap: SEV requires q35 but I think that is only loosely related. The main point here is that virConnectGetDomainCapabilities was designed to be invoked just before booting an instance of a specific arch / machine type, not during initialisation of a hypervisor host which can potentially host multiple archs / machine types | |
| 10:37:02 | aspiers | kashyap: but no rush, I will wait patiently ;-) thanks again! | |
| 10:37:05 | stephenfin | sean-k-mooney: If you want, go for it. I'm reviewing ralonsoh's native OVSDB patch atm | |
| 10:37:46 | kashyap | aspiers: Right. One more (couldn't resist), on getDomainCapabilities(), I mentioned it in this WIP spec too (for OVMF): https://review.openstack.org/#/c/506720/6/specs/rocky/approved/ovmf-with-secure-boot-and-smm.rst | |
| 10:38:04 | aspiers | :) | |
| 10:38:33 | sean-k-mooney | stephenfin: ok, cool let me know if you want me to respin the py3-functional patches. we might leave the +w of rodolfos change to jaypies | |
| 10:38:55 | sean-k-mooney | *to jaypipes | |
| 10:39:05 | stephenfin | yup, that's that my plan | |
| 10:52:05 | stephenfin | sean-k-mooney: Oh yeah, and no need to respin, though a follow-up would be nice | |
| 11:16:20 | sean-k-mooney | stephenfin: by the way the reason im making the vscode change is so i can run the functional test under a debugger by addign a break piont and clicking on just the test i want to run :) | |
| 11:18:09 | sean-k-mooney | so vim,emacs,nano is fine too and i use them form most of my work but having a real debugger makes writing new tests and validteing them much simpler | |
| 11:18:49 | sean-k-mooney | i have done this in the past with nova but its on my todo list to set it up again and then document it in the nova docs | |
| 11:25:08 | openstackgerrit | Yury Kulazhenkov proposed openstack/nova master: remove deprecated os_brick import from ScaleIO driver https://review.openstack.org/638592 | |
| 11:51:43 | sean-k-mooney | stephenfin: im going to resping the last sriov migration patch. adrianc didnt cherry pick my v14 change so im going to do that and then check if anything else form your comments still needs to be fixed. | |
| 12:15:12 | frickler | sean-k-mooney: regarding your comment on https://bugs.launchpad.net/bugs/1801919, we don't use "Fix Committed" anymore, see also https://docs.openstack.org/project-team-guide/bugs.html , though I don't remember the exact reasoning. but it has been this way for a couple of years now I think | |
| 12:15:14 | openstack | Launchpad bug 1801919 in devstack "brctl is obsolete use ip" [Undecided,In progress] - Assigned to Nate Johnston (nate-johnston) | |
| 12:15:52 | sean-k-mooney | frickler: that is quite missleading for a casual on looker | |
| 12:16:32 | sean-k-mooney | frickler: if we were to drop one of the two i woudl have dropped fix released | |
| 12:17:25 | frickler | sean-k-mooney: I think it doesn't work that way round, because the bug then would still show up as open in lp | |
| 12:17:29 | sean-k-mooney | frickler: i have had people email me in the past about bug that were marked as fix released when they were not | |
| 12:18:03 | sean-k-mooney | frickler: they should still show up as open | |
| 12:18:59 | sean-k-mooney | the reason i recived email before were because people were trying to deploy with the latest release and they were still hitting the bug because the isse was merged on master be we had not done a release | |
| 12:20:49 | sean-k-mooney | frickler: anyway if its not used anymore ill stop setting it but i thinik the current useage is flawed | |
| 12:21:36 | phasespace | when running "openstack hypervisor list", where does it retrieve the hostname of the hypervisor from? Is it stored in a particular table? | |
| 12:21:48 | frickler | sean-k-mooney: probably the release team would be to correct addressee if you wanted to continued that discussion, I'm just the messenger ;) | |
| 12:22:55 | sean-k-mooney | frickler: i might raise it at the fourm/ptg. it just makes my life harder to determin if things are fixed upstream or not when determining if we need to backport things downstream or not | |
| 12:23:14 | sean-k-mooney | frickler: thanks for letting me know | |
| 12:23:47 | sean-k-mooney | frickler: we proably need to add a new status to storyboard too by the looks of it as it does not track the released status either | |
| 12:23:52 | adrianc | sean-k-mooney: sorry for that, i didnt notice i was working on an older PS for the auto detach/attach of the sriov port commit | |
| 12:24:03 | sean-k-mooney | that said a tag could work too | |
| 12:24:14 | sean-k-mooney | adrianc: no worries | |
| 12:24:22 | sean-k-mooney | adrianc: just fixing it up now | |
| 12:24:34 | adrianc | thx ! | |
| 12:25:50 | sean-k-mooney | adrianc: by the way did you see that https://review.openstack.org/#/c/620123/ was being blocked by https://review.openstack.org/#/c/639074/ | |
| 12:26:10 | sean-k-mooney | ill kick off a recheck now that it has merged | |
| 12:26:34 | adrianc | yea, saw you mentioning that in the neutron IRC | |
| 12:27:45 | adrianc | was waiting for this to get pushed :) as it will conclude the neutron change for SR-IOV live migration | |
| 12:29:25 | sean-k-mooney | adrianc: ya so the recheck should be running now. you might need to rebase but it sould just do a merge with master and pick up the change | |
| 12:34:48 | adrianc | ack ill keep an eye on it sean-k-mooney | |
| 12:53:57 | mriedem | tssurya: i added you to https://review.openstack.org/#/c/621474/ because we could return server group info for servers in a down cell but that hasn't been implemented in that change, just wondering what you think | |
| 12:54:34 | mriedem | https://review.openstack.org/#/c/621474/23/nova/api/openstack/compute/views/servers.py@203 | |
| 12:56:20 | bauzas | efried: when you're up, wanted to discuss with you about https://review.openstack.org/#/c/599208/15 | |
| 13:05:59 | tssurya | mriedem: looking | |
| 13:09:59 | aspiers | efried: thanks a lot for the helpful comments. New version up for review https://review.openstack.org/#/c/538498/17 | |
| 13:10:08 | aspiers | looks like reviewbot is AWOL BTW? | |
| 13:14:58 | tssurya | mriedem: its doable, and since we are doing a new microversion to add that we might as well include it into the partial response | |
| 13:16:06 | gmann | mriedem: do you know anything not backported/setup in live migration setup on devstack side for queens/pike ? | |
| 13:16:35 | gmann | mriedem: i am trying to enable the tempest-multinode-full job for stable branch and all migration test fail - http://logs.openstack.org/92/623992/1/check/tempest-multinode-full/45ca539/testr_results.html.gz | |
| 13:16:59 | gmann | work fine in rocky but fail in queens and pike | |
| 13:18:28 | gmann | seems sudo/public key issue ? - http://logs.openstack.org/92/623992/1/check/tempest-multinode-full/45ca539/controller/logs/screen-n-cpu.txt.gz#_Feb_25_14_59_15_559635 | |
| 13:22:57 | mriedem | gmann: i see that job is non-voting, so does it work otherwise for nova stable/queens changes? | |
| 13:23:35 | kashyap | aspiers: You here? Okay, so I just read the scrollback you linked. (And skimmed the patches.) | |
| 13:23:43 | gmann | mriedem: not yet as tempest restrict it to master only. | |
| 13:23:50 | aspiers | kashyap: here but was just about to go for a run in the sun :) | |
| 13:24:00 | kashyap | aspiers: Go for it. More important than staring at the screen. | |
| 13:24:16 | aspiers | kashyap: hehe glad we agree on that! biab | |
| 13:24:30 | kashyap | aspiers: Quick note is that let's not hardcode anything with: virConnectGetDomainCapabilities() | |
| 13:24:36 | kashyap | (Because we'll use it for other purposes.) | |
| 13:24:38 | gmann | mriedem: next question is should we make it enable and work for those stable branch or is it ok to run them from stable/rocky onwards ? | |
| 13:24:39 | kashyap | But let's talk later. | |
| 13:24:46 | aspiers | kashyap: you mean like hardcoding the arch or machine type? | |
| 13:24:50 | aspiers | or emulatorbin? | |
| 13:25:03 | mriedem | gmann: tempest-multinode-full or tempest-slow? that patch seems to be dealing with enabling tempest-slow on stable branches | |
| 13:25:17 | kashyap | aspiers: Machine type can _probably_ be hard-coded: as only Q35 supports it? | |
| 13:25:17 | gmann | mriedem: yes boih. | |
| 13:25:20 | gmann | both | |
| 13:25:36 | aspiers | kashyap: I was considering changing my patch so that the memoized _domaincaps is a dict of dicts mapping arch -> machine type -> caps XML | |
| 13:25:54 | aspiers | or yeah, maybe just arch -> caps | |
| 13:25:56 | mriedem | gmann: well we have the nova-live-migration job on stable so live migration must work | |
| 13:26:19 | kashyap | aspiers: Yeah, meonized _domaincaps sounds like a good start to me; but I want to think this through a bit more. | |
| 13:26:45 | aspiers | kashyap: emulatorbin can be obtained programmatically from the virConnectGetCapabilities() data we already have | |
| 13:26:53 | kashyap | I haven't had time to soak into the details yet. But I just wanted to make a quick note that for Secure Boot, we'll be using the getDomainCapabilities() | |
| 13:26:58 | aspiers | kashyap: it's already memoized, but only the results from a single API call | |
| 13:27:05 | kashyap | aspiers: Yep | |
| 13:27:15 | mriedem | gmann: i'm not sure what's going on with this http://logs.openstack.org/92/623992/1/check/tempest-multinode-full/45ca539/controller/logs/screen-n-cpu.txt.gz?level=TRACE#_Feb_25_14_57_07_488187 | |
| 13:27:25 | aspiers | kashyap: yep, I commented on your q35 spec noting the potential for reuse | |
| 13:27:27 | kashyap | aspiers: Also, on your `virsh` command from the scrollback, for non-Q35, I _do_ get "sev supported = no" on my Fedora-28: | |
| 13:27:30 | kashyap | $> virsh domcapabilities --virttype kvm --emulatorbin /usr/bin/qemu-kvm --arch x86_64 --machine pc-i440fx-1.4 | grep "sev supported" <sev supported='no'/> | |
| 13:27:46 | aspiers | that's what I would expect | |
| 13:27:48 | kashyap | aspiers: Thanks; I need to re-spin it 'soon' | |
| 13:27:58 | aspiers | :) | |
| 13:28:01 | mriedem | gmann: ssh setup between the compute nodes must be messed up | |
| 13:28:03 | aspiers | OK, back in ~1 hour | |
| 13:28:07 | kashyap | Sure. | |
| 13:28:14 | aspiers | maybe a bit more, since I'll also need lunch :) | |
| 13:28:20 | aspiers | and it's *very* sunny here ... | |
| 13:28:33 | kashyap | aspiers: I'm here for ~2-ish hours, then on flaky internet connections on trains | |
| 13:28:42 | aspiers | kashyap: OK | |
| 13:28:52 | aspiers | kashyap: BTW if you need more context / detail: https://specs.openstack.org/openstack/nova-specs/specs/stein/approved/amd-sev-libvirt-support.html | |
| 13:28:59 | kashyap | But we're in the same timezone, we can work it out. | |
| 13:29:09 | aspiers | ok cool | |
| 13:29:18 | kashyap | aspiers: Yeah, I saw it whoosh by in the past; just didn't have time to dig into it. | |