| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2019-02-20 | |||
| 19:01:57 | melwitt | you definitely need to enable the workaround if you're doing a live upgrade | |
| 19:02:27 | imacdonn | yeah, seems we've been here before ;) At least removing the service seems to actually not generally break things this time ;) | |
| 19:02:35 | aspiers | melwitt: are release notes only for operator- / user-facing changes? | |
| 19:02:54 | aspiers | just wondering if I should add one for this new driver-capabilities-to-traits patch | |
| 19:06:42 | imacdonn | heh | |
| 19:07:07 | melwitt | aspiers: yeah, anything users should know about a change. I don't think I've seen the driver capabilities to traits thing yet | |
| 19:07:29 | aspiers | melwitt: https://review.openstack.org/#/c/538498/12 | |
| 19:07:40 | aspiers | melwitt: it's my rebase of an old patch from mriedem | |
| 19:07:45 | aspiers | I'm currently fixing the test failures | |
| 19:09:31 | imacdonn | aspiers: I guess my question (if I mattered) would be: who needs to know about this ? | |
| 19:09:45 | melwitt | aspiers: ok, I see. I think that would use a release note to let users know that driver capabilities will show up as custom traits in the placement API | |
| 19:10:18 | aspiers | imacdonn: my question too ;-) | |
| 19:10:24 | aspiers | melwitt: OK thanks! | |
| 19:12:26 | melwitt | aspiers: as for who, I don't know exactly, but it's the kind of thing I could see a cloud monitoring system looking at, what driver capabilities exist on each compute host | |
| 19:12:51 | aspiers | melwitt: yup, makes sense | |
| 19:13:22 | imacdonn | if there's a change to the way things appear through any published interface, I'd say it definitely should be a release-note. | |
| 19:13:52 | aspiers | yeah, I suppose just new traits appearing would be sufficient to justify documenting | |
| 19:18:15 | openstack | Launchpad bug 1816831 in OpenStack Compute (nova) "DOC: typo in add_initial_allocation_ratio releasenote" [Undecided,New] | |
| 19:18:15 | imacdonn | tangentially related (to release notes), I found a typo in another one. I might submit a fix a bit later, if I get time, but if someone else wants to grab it, that'd be cool too. https://bugs.launchpad.net/nova/+bug/1816831 | |
| 19:20:52 | melwitt | thanks | |
| 19:33:23 | mriedem | dansmith: +2 on that docs change | |
| 19:33:58 | dansmith | mriedem: +W | |
| 19:34:05 | mriedem | let us never speak of this again | |
| 19:34:25 | dansmith | I looked a the best practices one a bit, found one typo, but will look again when it gets respun | |
| 19:35:42 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Fix typo in initial_disk_allocation_ratio release note https://review.openstack.org/638245 | |
| 19:35:44 | mriedem | imacdonn: done - thanks for reporting that ^ | |
| 19:36:10 | imacdonn | mriedem: thanks! | |
| 19:37:12 | dansmith | +W within one minute | |
| 19:37:46 | melwitt | whew, what a sense of accomplishment | |
| 19:39:28 | aspiers | almost as towering an achievement as https://review.openstack.org/#/c/609777/ | |
| 19:39:45 | melwitt | lol | |
| 19:39:49 | openstackgerrit | Matt Riedemann proposed openstack/python-novaclient master: Add support for microversion 2.70 - expose device tags https://review.openstack.org/636779 | |
| 19:39:54 | aspiers | melwitt: my finest hour | |
| 19:46:04 | artom | aspiers, I'm grateful that your around | |
| 19:48:23 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Replace ansible --sudo with --become in live_migration/hooks scripts https://review.openstack.org/635308 | |
| 19:52:41 | imacdonn | artom: I see what you did there ;) | |
| 20:02:27 | artom | imacdonn, I no you did. | |
| 20:02:36 | imacdonn | :) | |
| 20:03:12 | aspiers | artom: stop teasing, its cruel | |
| 20:18:01 | aspiers | ;) | |
| 20:52:41 | mriedem | low-hanging-fruit docs bug if someone wants it https://bugs.launchpad.net/nova/+bug/1816859 | |
| 20:52:42 | openstack | Launchpad bug 1816859 in OpenStack Compute (nova) "Server concepts in nova - automatic resize confirm is wrong in docs" [Medium,Triaged] | |
| 20:54:30 | openstackgerrit | Adam Spiers proposed openstack/nova master: Convert driver supported capabilities to compute node provider traits https://review.openstack.org/538498 | |
| 20:59:09 | openstackgerrit | Adam Spiers proposed openstack/nova master: Convert driver supported capabilities to compute node provider traits https://review.openstack.org/538498 | |
| 21:00:01 | aspiers | efried_pto, mriedem, melwitt: I think this is ready for serious consideration now ^^^ | |
| 21:00:24 | aspiers | I've fixed all the TODOs I can see, and (fingers crossed) the tests will pass | |
| 21:00:35 | aspiers | s/will/should/ | |
| 21:04:04 | melwitt | kewl | |
| 21:08:24 | mriedem | aspiers: so is your sev stuff going to rebase on top of that or what? | |
| 21:09:09 | mriedem | because it refers to both a blueprint that's already complete and one that's not approved | |
| 21:15:19 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Add cross-cell resize policy rule and enable in API https://review.openstack.org/638269 | |
| 21:15:19 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Confirm cross-cell resize while deleting a server https://review.openstack.org/638268 | |
| 21:15:26 | mriedem | dansmith: ^ that's the end of the road, let's ship it | |
| 21:34:33 | openstackgerrit | Sundar Nadathur proposed openstack/nova master: ksa auth conf and client for cyborg access https://review.openstack.org/631242 | |
| 21:36:04 | artom | mriedem, heads up, we're hitting something similar to https://bugs.launchpad.net/nova/+bug/1813789 in our CI, except with revert_resize rather than evacuate | |
| 21:36:05 | openstack | Launchpad bug 1813789 in OpenStack Compute (nova) "Evacuate test intermittently fails with network-vif-plugged timeout exception" [Medium,Confirmed] | |
| 21:36:16 | artom | Ie, listening too late for an event | |
| 21:36:27 | artom | So looks like I'm fixing both of those | |
| 21:37:01 | artom | Would it make sense to file a different bug? Or just append to that one? | |
| 21:37:31 | sean-k-mooney | i would just use ^ | |
| 21:37:52 | mriedem | artom: that bug is f'ed | |
| 21:38:03 | mriedem | i think we *are* registered to wait for the event | |
| 21:38:05 | mriedem | and neutron isn't sending it | |
| 21:38:34 | sean-k-mooney | mriedem: in the down stream bug neutron is sending it before we register to recive it | |
| 21:38:48 | mriedem | need to figure out what is triggering the event then | |
| 21:38:50 | mriedem | because for evacuate, | |
| 21:38:54 | artom | mriedem, and that's what you describe in your bug as well | |
| 21:39:00 | mriedem | the driver.spawn() is registering the callback | |
| 21:39:14 | mriedem | what might be sending the event is the port binding change | |
| 21:39:36 | mriedem | does that happen before driver.spawn (evacuate) and driver.finish_revert_migration? | |
| 21:40:16 | sean-k-mooney | mriedem: this is the downstream bug by the way https://bugzilla.redhat.com/show_bug.cgi?id=1678681 | |
| 21:40:18 | openstack | bugzilla.redhat.com bug 1678681 in openstack-nova "REVERT_RESIZE stuck for 300s: "VirtualInterfaceCreateException: Virtual Interface creation failed" [Medium,Assigned] - Assigned to alifshit | |
| 21:40:23 | artom | mriedem, so, would it make sense to start listening immediately after calling Neutron? | |
| 21:40:41 | sean-k-mooney | artom: i would do it just before | |
| 21:40:53 | artom | sean-k-mooney, heh, right, that makes more sense | |
| 21:40:56 | mriedem | you have to register the callback before | |
| 21:41:24 | artom | Hrmm, and we might not have the callback until we get to spawn()? | |
| 21:41:29 | mriedem | sec | |
| 21:42:28 | mriedem | artom: so for evacuate, this is what updates the port binding to point at the dest host https://github.com/openstack/nova/blob/master/nova/compute/manager.py#L3212 | |
| 21:42:42 | mriedem | the driver spawn then happens here https://github.com/openstack/nova/blob/master/nova/compute/manager.py#L3276 | |
| 21:43:06 | mriedem | so neutron likely doesn't send a 2nd network-vif-plugged event because it's already sending one for the port binding change | |
| 21:43:08 | mriedem | that's my guess anyway | |
| 21:43:31 | artom | (mriedem, btw, I sill never stop being impressed by the speed at which you walk through code and paste the relevant bits in here) | |
| 21:43:41 | mriedem | as for the revert resize, | |
| 21:43:46 | mriedem | port binding changes back to the source here https://github.com/openstack/nova/blob/master/nova/compute/manager.py#L4089 | |
| 21:43:59 | mriedem | and the guest is re-spawned (and vifs are plugged) here: https://github.com/openstack/nova/blob/master/nova/compute/manager.py#L4108 | |
| 21:44:10 | mriedem | i'm fast on *this* because i've been in the bowels of it for 6 months | |
| 21:45:02 | artom | Superbowels | |
| 21:45:10 | artom | OK, my eyes are glazing over. | |
| 21:45:17 | artom | Will pick up tomorrow. | |
| 21:45:17 | sean-k-mooney | os with ml2 ovs. the network-vif-plugged events are sent by the nova ovs agent when it finishes wiring up the port | |
| 21:45:36 | sean-k-mooney | if we had not got to the point of cleaning up the ports on the souce node | |
| 21:45:42 | artom | mriedem's link will be in my IRC log for future consultation. | |
| 21:46:11 | sean-k-mooney | when we rebind to the source node it might send the network-vif-plugged event imiddately | |
| 21:46:38 | mriedem | artom: no need https://bugs.launchpad.net/nova/+bug/1813789/comments/2 | |
| 21:46:40 | openstack | Launchpad bug 1813789 in OpenStack Compute (nova) "Evacuate test intermittently fails with network-vif-plugged timeout exception" [Medium,Confirmed] | |
| 21:46:52 | mriedem | i put the notes in the bug | |
| 21:47:29 | artom | mriedem, cheers! | |
| 21:50:08 | mriedem | we also have the same issue in unshelve https://github.com/openstack/nova/blob/a4e6340bdcc2fda8fed41abc64ea554d9a8ab713/nova/compute/manager.py#L5087 | |
| 21:51:51 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Add cross-cell resize policy rule and enable in API https://review.openstack.org/638269 | |