| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2019-11-06 | |||
| 14:30:18 | sean-k-mooney | oh... | |
| 14:30:29 | sean-k-mooney | yes you could do that | |
| 14:30:52 | sean-k-mooney | but its passing form me because its using the cached wheel | |
| 14:31:04 | sean-k-mooney | if i try to use 5.5.0 i get your error | |
| 14:31:15 | efried | like you already had 5.7.0 installed before you started? | |
| 14:31:18 | sean-k-mooney | actully i get | |
| 14:31:20 | sean-k-mooney | ERROR: failed virConnectSetIdentity | |
| 14:31:22 | sean-k-mooney | ERROR: failed virDomainGetGuestInfo | |
| 14:31:24 | sean-k-mooney | error: command '/usr/bin/python' failed with exit status 1 | |
| 14:31:26 | efried | yes, that's what I'm seeing. | |
| 14:31:40 | efried | it's slightly different with 5.5 vs 5.7 | |
| 14:32:02 | sean-k-mooney | so i proably had libvirt python 5.7 alredy install as that is the lates version | |
| 14:32:08 | sean-k-mooney | and i had previously stacked | |
| 14:32:20 | sean-k-mooney | ill ask downstream one sec | |
| 14:35:23 | sean-k-mooney | efried: it might be this? https://github.com/libvirt/libvirt-python/commit/873e0ca7db3ae43bd514354997cdfa7ea5b7f70a | |
| 14:36:23 | efried | you mean that's the fix for same | |
| 14:36:30 | efried | kinda looks that way, yeah. | |
| 14:36:44 | efried | It wouldn't surprise me that libvirt-python's version has to match that of libvirt itself. | |
| 14:36:56 | efried | lemme clean and try building 5.7.0 | |
| 14:38:19 | sean-k-mooney | efried: so apparently the libvirt python version needs to be newer or the same as the libvir tyou are compiling | |
| 14:38:22 | efried | I guess libvirt-python hasn't been published to pypi yet or something? Why wouldn't `pip install -U` pick up the latest? | |
| 14:39:19 | efried | tried asking explicitly for 5.8.0 and it bounces | |
| 14:39:49 | sean-k-mooney | ya let me check pypy you can install it from souce if you need too | |
| 14:40:06 | efried | nah, I'll just bust down to libvirt 5.7.0 | |
| 14:40:10 | efried | I only need 5.6.0 min | |
| 14:41:14 | efried | but I guess an improvement for your script would be a version qualifier on libvirt-python >= whatever version of libvirt you're compiling. | |
| 14:41:44 | sean-k-mooney | yep | |
| 14:41:45 | efried | I don't know if that failure ("can't find libvirt-python 5.8.0") would be more obvious or not. | |
| 14:43:00 | sean-k-mooney | i proably can also do a safty check or install it form souce. | |
| 14:43:13 | sean-k-mooney | it would not be hard to just clone it and check out the same tag | |
| 14:43:16 | sean-k-mooney | then pip install it | |
| 14:43:49 | efried | depends what the plugin user wants it for. If it's not available on pypi that means it's not released yet, right? | |
| 14:44:11 | efried | Though I thought tags don't move once they're set. | |
| 14:44:38 | sean-k-mooney | efried: no its released just talking to danpb he though it was released he is fixing it now | |
| 14:44:48 | efried | o | |
| 14:45:36 | sean-k-mooney | efried: its there now | |
| 14:46:42 | efried | cool. I've already got both of these stacking at 5.7.0; if they fail I'll bump one up. | |
| 14:47:04 | efried | I guess now we'll see if libvirt-python 5.8.0 works with libvirt 5.7.0 :P | |
| 14:48:29 | sean-k-mooney | i manulaly installed it and it did at least on my host | |
| 14:52:16 | sean-k-mooney | melwitt: by the way have you seen migrate.exceptions.VersionNotFoundError: \"Database schema file with version 72 doesn't exist.\" before | |
| 14:52:35 | sean-k-mooney | melwitt: were you looking at somthing similar a while ago | |
| 14:53:00 | sean-k-mooney | that is form a "nova-manage api_db sync" | |
| 15:16:07 | mriedem | gibi_ptg: when it's not past 11pm in china and you're free, you were +2 on this cross cell change https://review.opendev.org/#/c/635668/ and now dan is too after some updates if you can hit it again | |
| 15:24:09 | sean-k-mooney | mriedem: you generally know what direction to look to root casue thing with db migrations | |
| 15:24:18 | sean-k-mooney | mriedem: does http://paste.openstack.org/show/785852/ look familar | |
| 15:24:33 | sean-k-mooney | igrate.exceptions.VersionNotFoundError: \"Database schema file with version 72 doesn't exist.\"", | |
| 15:24:55 | sean-k-mooney | im not sure where the version 72 comes form in the code | |
| 15:26:01 | sean-k-mooney | oh its here https://github.com/openstack/nova/tree/master/nova/db/sqlalchemy/api_migrations/migrate_repo/versions | |
| 15:38:06 | dansmith | sean-k-mooney: yeah, those are where the indexes are | |
| 15:38:30 | dansmith | sean-k-mooney: usually that error happens when you have a pyc of the migration but not the py of it, because you have a dirty working directory or something | |
| 15:39:01 | sean-k-mooney | dansmith: in this case its becaue i deploy kolla with master after you added the placeholders | |
| 15:39:17 | sean-k-mooney | but the current master contaienr are now clamped to traing | |
| 15:39:23 | sean-k-mooney | *train | |
| 15:39:32 | sean-k-mooney | while they are preparing an rc | |
| 15:39:43 | sean-k-mooney | so the api sysnc is trying to do a downgrade | |
| 15:39:52 | dansmith | sure, but.. the only thing that causes it to know about migration 72 is the fact that that file exists on disk | |
| 15:39:55 | sean-k-mooney | which obviously wont work | |
| 15:40:03 | dansmith | downgrade? | |
| 15:40:27 | sean-k-mooney | they dont exist on stable train right | |
| 15:40:30 | dansmith | that shouldn't cause that error | |
| 15:40:49 | dansmith | unless you have a database with version 72 and are trying to run downgrade I guess | |
| 15:40:56 | mriedem | downgrade isn't around anymore | |
| 15:41:19 | mriedem | sdague ripped it out a long time ago | |
| 15:41:25 | dansmith | mriedem: it's just a pass and technically runs IIRC | |
| 15:41:25 | sean-k-mooney | ya | |
| 15:41:36 | sean-k-mooney | its fine | |
| 15:41:38 | mriedem | the method doesn't exist in the scripts anymore | |
| 15:41:42 | sean-k-mooney | its just my home openstack | |
| 15:41:43 | dansmith | but I dunno why anything would be trying to do it | |
| 15:42:02 | dansmith | mriedem: we know that | |
| 15:42:03 | sean-k-mooney | it was bad timing that i deploy master just after nova rc1 when that was merged | |
| 15:42:29 | dansmith | sean-k-mooney: so yeah must be that sqla-migrate sees the db version as 72 and is looking for the downgrade script I guess? but..that'd be your problem | |
| 15:42:38 | sean-k-mooney | but the current "master" kolla image are stable train as they have not done there rc1 yet | |
| 15:43:18 | sean-k-mooney | ya i can just build the container form master nova and continue or redeploy with stable train | |
| 15:43:18 | mriedem | heh so you time traveled | |
| 15:43:33 | sean-k-mooney | its not a big deal | |
| 15:44:50 | openstackgerrit | Matt Riedemann proposed openstack/nova stable/stein: Add --dry-run option to heal_allocations CLI https://review.opendev.org/693199 | |
| 15:44:51 | openstackgerrit | Matt Riedemann proposed openstack/nova stable/stein: Add BFV wrinkle to TestNovaManagePlacementHealAllocations https://review.opendev.org/693201 | |
| 15:44:51 | openstackgerrit | Matt Riedemann proposed openstack/nova stable/stein: Add --instance option to heal_allocations https://review.opendev.org/693200 | |
| 15:46:28 | sean-k-mooney | well now i know where to look and how that works. im off to lunch brb | |
| 15:53:27 | openstackgerrit | Boris Bobrov proposed openstack/nova master: Also enable iommu for virtio controllers and video in libvirt https://review.opendev.org/684825 | |
| 15:53:28 | openstackgerrit | Boris Bobrov proposed openstack/nova master: Create a controller for qga when SEV is used https://review.opendev.org/693072 | |
| 16:04:06 | openstackgerrit | Boris Bobrov proposed openstack/nova master: Move rng device checks to the appropriate method https://review.opendev.org/693210 | |
| 16:07:13 | efried | Nov 06 09:22:55 breakfast dnsmasq[27199]: exiting on receipt of SIGTERM | |
| 16:07:13 | efried | Nov 06 09:19:25 breakfast dnsmasq-dhcp[27199]: read /var/lib/libvirt/dnsmasq/default.hostsfile | |
| 16:07:13 | efried | The libvirt-bin service is inactive. There's no journal, but systemctl shows these as the last two lines: | |
| 16:07:13 | efried | '/var/run/libvirt/libvirt-sock': No such file or directory. | |
| 16:07:13 | efried | sean-k-mooney: back to n-cpu failing to start because of | |
| 16:07:14 | efried | I don't see anything in the stack logs that looks like it ought to have shut the service down. | |
| 16:08:22 | efried | On both stacks the systemctl status shows that ~3.5min gap between loading default.hostsfile and SIGTERM | |
| 16:13:43 | efried | sean-k-mooney: I was able to manually start libvirt-bin and it hasn't stopped itself, so clearly something else was hitting it... | |
| 16:14:50 | efried | ...but with the service running, n-cpu still fails to come up in the same way. | |
| 16:21:40 | openstackgerrit | Boris Bobrov proposed openstack/nova master: Create a controller for qga when SEV is used https://review.opendev.org/693072 | |
| 16:21:40 | openstackgerrit | Boris Bobrov proposed openstack/nova master: Also enable iommu for virtio controllers and video in libvirt https://review.opendev.org/684825 | |
| 16:22:26 | mriedem | amodi: melwitt: looks like we can keep the iso file without changes https://review.opendev.org/#/c/693065/7 | |
| 16:22:41 | mriedem | it was already fixed in the manuals repo before i imported that doc https://github.com/openstack/openstack-manuals/commit/1206b9b727594ae04f0d2a4818f183e95253b464 | |
| 16:22:43 | openstackgerrit | Merged openstack/nova master: Pass exception through TaskBase.rollback https://review.opendev.org/692689 | |
| 16:25:55 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Revert "openstack server create" to "nova boot" in nova docs https://review.opendev.org/693065 | |
| 16:27:00 | mriedem | this is why i'm not as free willy on accepting any old docs change since it's a pain in the butt to fix them if they are wrong | |
| 16:30:06 | melwitt | haha free willy | |