Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2020-10-06
10:27:37 sean-k-mooney shoudl this not be in the ceph devstack plugin repo instead
10:27:39 lyarwood sean-k-mooney: I'm just failing at sharing the list of generated keyrings between the controller and compute, it's a basic ansible thing I'm failing on
10:28:09 lyarwood sean-k-mooney: zuul wasn't happy when I started with it in the plugin
10:28:19 lyarwood sean-k-mooney: so I'm leaving it here until it's working
10:28:31 lyarwood gah baby woke up, brb
10:29:57 sean-k-mooney lyarwood: ok well you could alos look at https://github.com/openstack/devstack/tree/master/roles/sync-devstack-data
10:30:17 sean-k-mooney that copies the tls certs for multinode setups beteween hosts
10:32:16 sean-k-mooney its using the synchronize role instead of fetch and copy
10:47:06 lyarwood sean-k-mooney: yeah also looked at that in previous PS's and got confused so I switched back.
10:47:25 lyarwood sean-k-mooney: my main issue is with how I should store and share the list of keyrings between the hosts
10:47:58 lyarwood sean-k-mooney: hmm do I set it as a fact against the controller and then look that up from the compute?
10:48:06 sean-k-mooney lyarwood: you proably just want to sync all of them to the zuul executor and then all of them to the subnodes
10:48:38 lyarwood sean-k-mooney: right but I don't think any of the copy/sync modules support wildcards do they?
10:48:39 sean-k-mooney so just copy all of them into a singel folder and copy all the files in the folder
10:49:37 sean-k-mooney so you basically want to do https://github.com/openstack/devstack/blob/master/roles/sync-devstack-data/tasks/main.yaml#L21-L30 but replace the with_items
10:49:38 lyarwood k I'll give that a go once my wife gets home and I can throw the baby at her
10:49:56 lyarwood sean-k-mooney: right but the hardcoded list was something I wanted to avoid
10:50:04 sean-k-mooney ya you can
10:50:17 sean-k-mooney you can register the files as a fact on the contoler
10:50:27 sean-k-mooney then use it on the contoler to sync it to the exectutor
10:50:32 lyarwood yup
10:50:44 lyarwood the part I was missing was that it wasn't a local fact on the compute
10:50:46 sean-k-mooney then do the same thing on the executor to push them
10:50:58 lyarwood ansible is weird sometimesd
10:51:15 sean-k-mooney well its sort as a host var
10:51:27 sean-k-mooney most facts are specific to the host they are created on
10:51:57 sean-k-mooney but the point is to pull the data from the contoler to the executor then push it form the executor to the compute nodes
10:52:12 sean-k-mooney you never need to run the sync on the computes
10:53:02 sean-k-mooney if you know the contolers name you can also just use that to get its fact
11:10:39 openstackgerrit Rajat Dhasmana proposed openstack/nova master: WIP: Add support of blockCommit when VM is down https://review.opendev.org/756261
11:18:07 tosky lyarwood: I think sean-k-mooney already gave more information than I could have provided
11:24:13 openstackgerrit Elod Illes proposed openstack/nova stable/stein: Set different VirtualDevice.key https://review.opendev.org/749415
11:33:52 lyarwood tosky: ack thanks
11:41:37 andrewbonney Hi. We're hitting an issue with live migration after moving to Ussuri which I think may be down to nova not handling some output from libvirt. I've put some debug here if anyone could take a look: http://paste.openstack.org/show/798740/. I'd be happy to post that to launchpad if preferred
11:42:36 lyarwood bauzas / melwitt ; https://review.opendev.org/#/q/branch:%255Estable/.*+status:open+owner:%22Lee+Yarwood+%253Clyarwood%2540redhat.com%253E%22+label:Code-Review%253E%253D2 - btw my stable backlog is starting to build up if either of you have time this week
11:48:02 gibi andrewbonney: seems like a valid bug. Could you please file a bug report in launchpad for it https://bugs.launchpad.net/nova/+filebug
11:48:21 andrewbonney Will do, thanks
12:04:23 bauzas lyarwood: ack, will look
13:51:09 lyarwood bauzas: many thanks btw
13:52:35 bauzas lyarwood: np, took a bit to verify all the changes, but thanks for it
13:56:32 lyarwood fun we dump the devstack run from a controller into the job logs twice for multinode jobs at the moment
13:56:36 lyarwood 29871 2020-10-06 12:07:45.753810 | TASK [run-devstack : Run devstack]
13:56:39 lyarwood 29872 2020-10-06 11:42:53.973082 | controller | + unset GREP_OPTIONS
13:57:02 lyarwood ^ that's the compute run with logs from the previously already logged controller run
13:57:07 lyarwood /o\
14:33:35 openstackgerrit Lee Yarwood proposed openstack/nova master: libvirt: Support extending locally attached rbd volumes https://review.opendev.org/752069
16:05:30 openstackgerrit Ghanshyam Mann proposed openstack/nova master: DNM: testing tempest copy-image test https://review.opendev.org/742547
16:17:57 openstackgerrit Lee Yarwood proposed openstack/nova master: WIP: nova-next: Start testing the 'q35' machine type https://review.opendev.org/708701
18:14:56 sean-k-mooney melwitt: can you take a look at https://review.opendev.org/#/c/751642/6 or lyarwood if you are about its relitivly minor but i need it for the follow up patch
18:20:33 dansmith sean-k-mooney: your comment in driver.py is exactly why N->N+m is not just doable when rpc versions match :)
18:21:53 sean-k-mooney ah becasue we somethimes move things
18:22:03 sean-k-mooney without chainging rpc verions
18:22:18 dansmith you're doing it here, yeah
18:22:37 sean-k-mooney yep for what its worth im doing it for upgrade reasons
18:22:40 dansmith I'm not saying change it, because we do other stuff like it, I'm just pointing out this as an example
18:22:49 sean-k-mooney ya
18:23:03 dansmith just something to keep in mind the next time someone asks why :)
18:23:14 sean-k-mooney ah yep
18:26:52 dansmith sean-k-mooney: check my questions on that and tell me if I should be -1ing
18:28:57 sean-k-mooney looking now
18:32:38 openstackgerrit Merged openstack/nova stable/queens: compute: Allow snapshots to be created from PAUSED volume backed instances https://review.opendev.org/729178
18:32:47 openstackgerrit Merged openstack/nova stable/queens: Improve CinderFixtureNewAttachFlow https://review.opendev.org/749228
18:32:54 openstackgerrit Merged openstack/nova stable/queens: Robustify attachment tracking in CinderFixtureNewAttachFlow https://review.opendev.org/749229
18:35:50 sean-k-mooney dansmith: you can -1 i am missing the test. i dont need the host check but im not asserting we ignore and log the error in any of the tests
18:36:58 sean-k-mooney dansmith: i commented in line. is there a better way to handel the change in behavior. where im moving work that was prviously done in one api call to another?
18:37:21 sean-k-mooney its rare we do this but it sometimes happens
18:37:54 dansmith sean-k-mooney: no, I just wanted to note it, same as here
18:38:35 sean-k-mooney cool
18:41:48 sean-k-mooney hum same host resize. perhaps that is why we do that check.
18:42:07 dansmith sean-k-mooney: ... :)
18:42:18 sean-k-mooney ill chekc the call flow again when i add the unit test
18:42:24 dansmith cool
18:42:51 sean-k-mooney i think the new unplug call would be before the netowrkign is plugged for the dest
18:43:23 sean-k-mooney so i think it still works but ill make sure and either add the same check or leave a comment in the code
18:45:26 sean-k-mooney one of the side effect of when libvirt is in charge fo when libvirt is responcible for addign and removing port form ovs is
18:45:41 sean-k-mooney on same host resize it remvoe and add the port to ovs
18:45:56 sean-k-mooney even when we dont call unplug_vifs
18:46:56 sean-k-mooney e.g. if you using ovn or ovs with the ovs firewall then libvirt removes the ovs port when we destory the domain
18:47:56 sean-k-mooney plug_vif in os vif is a noop in that case. that is what the next patch changes
18:48:49 sean-k-mooney with https://review.opendev.org/#/c/602432/21 os-vif now does the plugging but i need the call to unplug_vifs eairler to have the set of events that neutron sends be the same in the same host resize case
18:51:20 openstackgerrit Navid Pustchi proposed openstack/nova-specs master: add availability-zone-affinity-anti-affinity-filter review. https://review.opendev.org/756380
18:55:30 openstackgerrit Navid Pustchi proposed openstack/nova-specs master: add availability-zone-affinity-anti-affinity-filter review. https://review.opendev.org/756380
20:44:13 rm_work can i see from the nova endpoint which version I am actually running? I assumed so? but the versions i'm seeing listed didn't change when i upgraded so i can't tell if my upgrade failed or if i'm just looking at the wrong spot
20:52:27 rm_work ok yeah nm, verified my upgrade failed :D
21:26:42 openstackgerrit Navid Pustchi proposed openstack/nova-specs master: add availability-zone-affinity-anti-affinity-filter review. https://review.opendev.org/756380
21:39:50 openstackgerrit Navid Pustchi proposed openstack/nova-specs master: add availability-zone-affinity-anti-affinity-filter review. https://review.opendev.org/756380
21:51:36 openstackgerrit melanie witt proposed openstack/nova stable/stein: Sanity check instance mapping during scheduling https://review.opendev.org/756404
21:58:07 openstackgerrit Moshe Levi proposed openstack/os-vif stable/victoria: Refactor code of linux_net to more cleaner and increase performace https://review.opendev.org/756406
23:01:29 openstackgerrit Merged openstack/nova master: Add Python3 wallaby unit tests https://review.opendev.org/754322
23:10:56 openstackgerrit melanie witt proposed openstack/nova master: Initialize global data separately and run_once in WSGI app init https://review.opendev.org/733627
#openstack-nova - 2020-10-07
02:38:12 openstackgerrit Wonil Choi proposed openstack/nova stable/rocky: Set instance host and drop migration under lock https://review.opendev.org/756421
05:51:11 openstackgerrit Ade Lee proposed openstack/nova master: Replace md5 with oslo version https://review.opendev.org/756434
07:08:42 gibi good morning Nova
08:31:49 openstackgerrit Jorhson Deng proposed openstack/nova master: add task state for attach volume and detach volume https://review.opendev.org/754258
08:36:35 kashyap gibi: If you're feeling security-inclined this morning, see if my questions here make sense :-) - https://review.opendev.org/#/c/756434/
08:39:17 gibi kashyap: looking
08:39:51 kashyap It's not from me, but it's a bulk change across components so that OpenStack works in a FIPS-based setup
08:40:31 kashyap Many moons ago /me worked on testing FIPS stuff at Red Hat :)
08:40:50 kashyap (https://en.wikipedia.org/wiki/Federal_Information_Processing_Standards)
08:41:20 kashyap (A better explanation is on Mozilla's blokg: https://developer.mozilla.org/en-US/docs/Mozilla/Projects/NSS/FIPS_Mode_-_an_explanation)

Earlier   Later