Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2022-06-10
07:59:30 jamespage_ tobias-urdin: hiya - I noticed your comment about re-creating mdevs prior to nova-compute startup on https://review.opendev.org/c/openstack/nova/+/838976 and was interested in what you might have todo that
08:00:12 jamespage_ I can see that the way forward is to use persistent mdevs in libvirt but I'm trying to figure out a sane way to dealing with host reboots with attached vgpus for older releases of openstack + libvirt
08:26:13 tobias-urdin jamespage_: i can drop you what we have, it's not pretty but maybe it can help you forward
08:26:40 jamespage_ tobias-urdin: yes please :)
08:27:24 tobias-urdin we have systemd unit Type=oneshot that has Before=nova-compute.service that executes this https://paste.opendev.org/show/bwReCtkRa4eiIYa64AqN/
08:27:32 tobias-urdin this is the remediate-dev code https://paste.opendev.org/show/bW91TyoqQCbFcPSTC3yT/
08:29:16 tobias-urdin there is some edge cases in that, and a little nasty undefine so you might want read through it and verify it for your use-case
08:31:42 jamespage_ tobias-urdin: ack thanks - I'll have a good review
09:05:28 opendevreview Rajesh Tailor proposed openstack/nova master: Remove return from rpc cast https://review.opendev.org/c/openstack/nova/+/845335
09:48:53 opendevreview Artom Lifshitz proposed openstack/nova stable/train: func: Add live migration rollback volume attachment tests https://review.opendev.org/c/openstack/nova/+/844745
09:48:54 opendevreview Artom Lifshitz proposed openstack/nova stable/train: fup: Assert state of connection_info during LM rollback in func tests https://review.opendev.org/c/openstack/nova/+/844746
11:45:33 opendevreview Rajesh Tailor proposed openstack/nova master: Remove return from rpc cast https://review.opendev.org/c/openstack/nova/+/845335
13:28:47 opendevreview Artom Lifshitz proposed openstack/nova stable/train: func: Add live migration rollback volume attachment tests https://review.opendev.org/c/openstack/nova/+/844745
13:28:48 opendevreview Artom Lifshitz proposed openstack/nova stable/train: fup: Assert state of connection_info during LM rollback in func tests https://review.opendev.org/c/openstack/nova/+/844746
14:37:49 opendevreview Artom Lifshitz proposed openstack/nova stable/train: fup: Assert state of connection_info during LM rollback in func tests https://review.opendev.org/c/openstack/nova/+/844746
14:42:42 sean-k-mooney dansmith: have you seen this before https://paste.opendev.org/show/bqi5a04gFzt0NE9SbHRV/
14:43:13 sean-k-mooney is this related to the perf stuff you were doing by anychance or is this unrelated
14:43:21 dansmith sean-k-mooney: well, during debugging, but that means we didn't install the plugin in early devstack
14:43:25 dansmith sean-k-mooney: yeah it's related
14:43:46 dansmith sean-k-mooney: we shouldn't configure it to use the plugin if we didn't install it.. is this in a job or local?
14:43:46 sean-k-mooney ok so i could either turn off the profileing or install it right
14:44:01 sean-k-mooney local with my ansible role stuff
14:44:19 dansmith yeah just set MYSQL_GATHER_PERFORMANCE=false
14:44:21 dansmith or whatever the variable is
14:44:30 sean-k-mooney i have not updted the verion of the devstack roles im using in a while so i can look into it
14:44:36 sean-k-mooney ok ill just disable it for now thanks
14:44:46 dansmith lib/databases/mysql: if [[ "$MYSQL_GATHER_PERFORMANCE" == "True" ]]; then
14:59:52 sean-k-mooney cool that worked
15:01:14 sean-k-mooney now i need to unstack and let ansible do it again to maksure it works end to end.
17:46:38 erlon hi nova folks, is there a policy for when tags are created for stable release branches?
18:50:50 melwitt erlon: the general policy is described here https://docs.openstack.org/project-team-guide/stable-branches.html#releases noting that branches in EM (Extended Maintenance) https://docs.openstack.org/project-team-guide/stable-branches.html#extended-maintenance will no longer be released
18:52:09 melwitt releases are requested by proposing a patch to the openstack/releases repo https://releases.openstack.org/reference/using.html#requesting-a-release and such patches must be acked by the PTL and/or release liasion (bauzas and elodilles)
18:52:26 melwitt these are examples for nova https://review.opendev.org/q/path:%255Edeliverables/.*/nova.yaml
18:58:09 erlon melwitt: so, the liason described in the policy is the requester (me in this case) of the poin release
18:58:13 erlon ?
18:59:33 melwitt erlon: no a liaison is a designated person other than the PTL who is able to ack release requests (the release team won't merge a release request until it is acked by the PTL or a release liasion)
19:01:22 erlon "The PTL or release liaison for a project may request a release from master by submitting a patch to this repository",
19:01:22 erlon The way is written, gives the impression that they need to submit the release patch, not only ACK and aproove
19:01:52 erlon so, IIC, I can submit the release request patch but they need to ACK and +1 it
19:02:12 erlon *IIUC
19:05:13 melwitt erlon: oh, I see. usually they are requested by a release liaison. I thought I had seen some where other people requested, maybe it was os-vif or other libs. to stay on the safer side just ping elodilles to let him know you want a new release for whatever branch and he'll let you know if the timing is ok and whatnot
19:06:19 erlon melwitt: right, thanks Mellany
19:06:50 melwitt np
19:06:55 erlon elodilles: ^ ?
19:09:33 melwitt erlon: he is likely not around right now due to timezone, I think he is UTC+1 or thereabouts. but he'll get back to you when he sees it
19:16:58 erlon hmm, ok, Ill talk to him monday
20:30:17 sean-k-mooney[m] others can request the release
20:30:53 sean-k-mooney[m] i have done it several times you just need to ping me or sylvain to +1 it
20:31:09 sean-k-mooney[m] erlon what repo did you want a release of?
20:36:49 erlon sean-k-mooney[m]: nova, for the xena and yoga releases
20:40:27 sean-k-mooney[m] ok ill try and take a look on monday to see what we have pending and we can disucss it in the meeting on tuesday. im currently the release liasons person for nova but i have mainly just looked at the libs rather then nova itself https://github.com/openstack/releases/blob/master/data/release_liaisons.yaml#L90
20:41:07 sean-k-mooney[m] the process is the same we just look at the pending pataches and what has been merged since the last release
20:42:26 sean-k-mooney[m] it there is nothing urgent or close to merging we can just use the current head fo the stable branch for the new release else we will wait a day or two for them to land and use the sha of with the patch we were waiting for
20:44:15 sean-k-mooney[m] most of the release process is automated there are sctips in the relases repo and zuul jobs to do the actual release and tagging
20:46:32 sean-k-mooney[m] https://github.com/openstack/releases/blob/master/tools/new_release.sh is the automated way to create the release patch
20:46:40 sean-k-mooney[m] but its not hard to do it by hand
20:48:27 sean-k-mooney[m] the next release for yoga would use be 25.0.1 https://github.com/openstack/releases/blob/master/deliverables/yoga/nova.yaml
21:24:02 melwitt sean-k-mooney[m]: urgh, sorry I missed you as liaison. I looked around trying to find it documented somewhere and failed. I will bookmark that page
21:28:42 opendevreview melanie witt proposed openstack/nova master: privsep: Move qemu-img create calls under nova.privsep.qemu https://review.opendev.org/c/openstack/nova/+/826750
#openstack-nova - 2022-06-11
00:17:12 opendevreview melanie witt proposed openstack/nova master: privsep: Move qemu-img create calls under nova.privsep.qemu https://review.opendev.org/c/openstack/nova/+/826750
#openstack-nova - 2022-06-13
02:22:14 opendevreview anguoming proposed openstack/placement master: Remove unicode literal strings https://review.opendev.org/c/openstack/placement/+/845531
05:51:45 opendevreview Takashi Kajinami proposed openstack/nova master: Retry attachment delete API call for 504 Gateway Timeout https://review.opendev.org/c/openstack/nova/+/845543
05:53:09 opendevreview Takashi Kajinami proposed openstack/nova master: Retry attachment delete API call for 504 Gateway Timeout https://review.opendev.org/c/openstack/nova/+/845543
05:55:14 opendevreview Takashi Kajinami proposed openstack/nova master: Retry attachment delete API call for 504 Gateway Timeout https://review.opendev.org/c/openstack/nova/+/845543
06:02:13 opendevreview Takashi Kajinami proposed openstack/nova master: Retry attachment delete API call for 504 Gateway Timeout https://review.opendev.org/c/openstack/nova/+/845543
06:04:03 opendevreview Takashi Kajinami proposed openstack/nova master: Retry attachment delete API call for 504 Gateway Timeout https://review.opendev.org/c/openstack/nova/+/845543
06:43:52 gibi gmann, dansmith: I missed the pings about the RBAC here from last week. Sorry for that I did not hang out on IRC
06:44:05 gibi overall I don't think we get much usefull feadback about RBAC
06:46:31 gibi there was one discussion that I can mention about the service role
06:47:33 gibi there it was expressed that some operators wants service specific service role
06:47:56 gibi like a role use only by cinder to talk to nova
06:48:11 gibi as this role could be defined pretty restrictively
06:48:28 gibi compared to a service role that is used by each service to talk to another service
06:48:50 gibi we agreed that this can be a later step in the service role work
06:50:36 gibi anther feedback here was that some operators want to define the a set of policy rules describing a service - service interaction from the client perspective
06:50:39 gibi e.g.
06:50:46 gibi cinder can define what action it wants to call on nova
06:50:57 gibi so the cinder tree there could be a policy file for that
06:51:11 gibi when cinder is installed
06:51:39 gibi that policy file needs to be copied to the policy.d of the nova
06:51:44 gibi to grant the access for cinder
08:31:56 gibi priteau: hi! thanks for the patch https://review.opendev.org/c/openstack/nova/+/845262 I'm +2 on it. Are you planning to backport this to stable branches too?
08:32:16 gibi bauzas: fyi ^^ a simple fix we agreed on last week in berlin ;)
08:32:49 opendevreview Takashi Kajinami proposed openstack/nova master: Retry attachment delete API call for 504 Gateway Timeout https://review.opendev.org/c/openstack/nova/+/845543
08:33:24 gibi sean-k-mooney: hi! Could you plug your +2 back to https://review.opendev.org/c/openstack/nova/+/829248 ? Thanks!
08:43:15 bauzas hello everyone
08:43:20 gibi Uggla: hi! I just started looking at your unshelve patch. Have you considered my comment about splitting it? https://review.opendev.org/c/openstack/nova/+/831507/11..13#message-9605cc5a2071898fa3da8d460761a3910f9bb55f
08:43:30 gibi bauzas: o/
08:43:36 bauzas gibi: hope you're good
08:43:59 gibi I'm on ~75% energy level
08:45:25 bauzas priteau: gibi: https://review.opendev.org/c/openstack/nova/+/845262 sent to the gate
08:45:31 gibi bauzas: thanks
08:45:42 bauzas agreed with you, we should backport it
08:45:43 gibi priteau: let me know your view on backporting. I can take it if you have not time for it
08:47:51 Uggla gibi, Hi. Yes I tried to split into 2 patches as you explained. But I had a lot of tests that failed. Mainly because az=None not handled correctly by the REST API part.
08:48:51 gibi Uggla: the first patch after the split should not alter the api behavior
08:49:10 Uggla gibi, trying to explain --> REST api and compute_api are "coupled". You need some code to handled the az=None case.
08:50:25 gibi ack. I think there should be a way to split it. But I haven't tried it so you have more info than me about the hardness of it
08:56:54 Uggla gibi, look at --> patchet 11 --> 12 is the attempt.
08:58:56 Uggla gibi, and the functional-py39 --> result.

Earlier   Later