Wednesday, 2022-10-19

opendevreviewJorhson Deng proposed openstack/nova master: Allow user defined stop type  https://review.opendev.org/c/openstack/nova/+/86133301:55
opendevreviewJorhson Deng proposed openstack/nova master: Allow user defined stop type  https://review.opendev.org/c/openstack/nova/+/86133302:17
opendevreviewJorhson Deng proposed openstack/nova master: Allow user defined stop type  https://review.opendev.org/c/openstack/nova/+/86133305:44
opendevreviewJorhson Deng proposed openstack/nova master: Allow user defined stop type  https://review.opendev.org/c/openstack/nova/+/86133305:54
*** bhagyashris is now known as bhagyashris|ruck06:24
sahido/07:53
sahidany chance that you have a second look at https://review.opendev.org/c/openstack/nova-specs/+/857838 sean-k-mooney? I hope to have addressed most of your comments I just have kept open one to ensure that this is addressed in the right way07:54
opendevreviewJorhson Deng proposed openstack/nova master: Allow user defined stop type  https://review.opendev.org/c/openstack/nova/+/86133308:02
opendevreviewAmit Uniyal proposed openstack/nova stable/zed: Adds a repoducer for post live migration fail  https://review.opendev.org/c/openstack/nova/+/86185609:21
opendevreviewAmit Uniyal proposed openstack/nova stable/zed: [compute] always set instance.host in post_livemigration  https://review.opendev.org/c/openstack/nova/+/86185709:21
mnasiadkaHello - any reason backport https://review.opendev.org/c/openstack/nova/+/829804 is hanging so long?10:25
sean-k-mooneymnasiadka: it just needs review it fell off my radar so i can take a look now10:28
mnasiadkasean-k-mooney: thanks!10:29
sean-k-mooneyin general if a patch is up and not reviewd for a period of tiem we expect the autorp or backporteer to ping us10:29
sean-k-mooneywe will eventualy get to it otherwise but the more vocal people are the more likely we are to see it10:29
sean-k-mooneyelodilles: if you have time can you take a look at https://review.opendev.org/c/openstack/nova/+/82980410:33
mnasiadkasean-k-mooney: Actually I need it in Wallaby, will do a cherry-pick today as well10:52
sean-k-mooneyack10:52
opendevreviewMichal Nasiadka proposed openstack/nova stable/wallaby: Fix to implement 'pack' or 'spread' VM's NUMA cells  https://review.opendev.org/c/openstack/nova/+/86183210:53
mnasiadkaactually it was clean, so simple as a click in Gerrit UI10:53
sean-k-mooneymnasiadka: you might also want the caching change we merged backported10:53
sean-k-mooneymnasiadka: gerrit wont set teh cherry-picked line correctly10:53
sean-k-mooneyit only has one10:54
sean-k-mooneyit should have one per branch10:54
sean-k-mooneyso gerrit only works the first time10:54
mnasiadkaok, I'll update10:55
mnasiadkasean-k-mooney: caching change? which one is that?10:55
opendevreviewAmit Uniyal proposed openstack/nova stable/yoga: Adds a repoducer for post live migration fail  https://review.opendev.org/c/openstack/nova/+/86187110:55
opendevreviewAmit Uniyal proposed openstack/nova stable/yoga: [compute] always set instance.host in post_livemigration  https://review.opendev.org/c/openstack/nova/+/86187210:55
sahidthank you sean-k-mooney[m], I have replied to one of you question, I'm not sure that we need to update the virt driver api but I may have missed something. At leaast the current impl that I have is working as expected for our usage11:30
sahidfor the other I will address them when i will get more reviews for sure11:30
sean-k-mooney1sahid: i belive you will need to chnage rebuild_instance to add the target state11:31
*** sean-k-mooney1 is now known as sean-k-mooney11:31
sean-k-mooneyso that the driver does not try to start the vm11:31
sahidI have reused the way that we currently handle evacuated instance that are stopped11:31
sahidthe driver start the instance and tjere is a process to stop it11:32
sean-k-mooneywe should not do that really11:32
sean-k-mooneysince the start is going to fail11:32
sean-k-mooneyat least in teh encypted volume case11:33
sean-k-mooneyideally you would pass the target state in and only start it if required11:33
sean-k-mooneywe can see what others think but if you say targetState=stop i would consider it a bug to attepmt to start it and then stop it11:34
sean-k-mooneyhttps://review.opendev.org/c/openstack/nova/+/858383/14/nova/compute/manager.py#3826 might be ok11:36
sean-k-mooneyno looking at the workflow11:37
sean-k-mooneyim pretty sure if you try that with an encypted volume the spawn will fail11:37
sean-k-mooneythe admin wont be abel to retirve the key form barbican11:38
sahidsean-k-mooney: oh i see, at least in our use-case it's fine as everything is done using cloud-init and we don't wait for that to happen12:04
sahidI'm wondering whether it's not mandotary to have the vm started to libvirt would have installed the tap and qemu sent the arp, we also know that ovs agent needs to see net devices to configure network12:06
sean-k-mooneythat woudl all happen when the user starts the vm12:07
sean-k-mooneywe shoudl perhaps discuss this in the ptg or at least get more eye on the review12:08
sean-k-mooneybinding the port in neutron makes sense im not sure that plugging the prot into ovs shoudl be done at this point12:08
sean-k-mooneywe need to prepare the disk image/instance direcotry12:09
sean-k-mooneybut im not sure we should go as far ad defining the domain12:09
sean-k-mooneywe could maintain the current beahivor of powering it off after but that does not fell correct to me12:10
sahidi also had this feeling when I have noticed it but then I felt like this was perhaps a necessity12:11
sahidi can make some tests but that will be difficult for me to test all the combinaisons 12:12
sahidfor the case that is open in your side, does that can be tested easily?12:12
sean-k-mooneyif we do that we need to catch the expction form looking up the key12:13
sean-k-mooneyand basically ignore it in this case and proceed to powering it off.12:13
sahidi see, i can for sure doing that, but as you mention we probably first have to see whether we want to avoid the vm to be started at driver side, which could brings others implications12:15
* sahid has some connectivity issues...12:24
opendevreviewribaudr proposed openstack/nova-specs master: Draft spec to support scaphandre  https://review.opendev.org/c/openstack/nova-specs/+/86188112:27
opendevreviewribaudr proposed openstack/nova-specs master: Draft spec to support scaphandre  https://review.opendev.org/c/openstack/nova-specs/+/86188112:38
bauzasreminder : our PTG sessions start in 5 mins, see you there in the bexar room12:55
bauzasstarted ^13:01
bauzasfolks, we need you !13:01
bauzaswe'll start with discussing about ELO'ing stable branches for stein and older13:05
bauzasnow*13:06
*** dasm|off is now known as dasm13:55
bauzaschateaulav: I know you're in the zoom room, but we are just discussing secure RBAC14:23
bauzasto sure you are listening14:23
bauzasnot* sure14:23
stephenfinsean-k-mooney: Your background blur feature is going nuts :)14:30
chateaulavbauzas: thanks14:30
sean-k-mooneyfixed by trunig it off14:39
gibivmaccel: we will discuss the PCI in placement stuff in https://www.openinfra.dev/ptg/rooms/bexar in 3 minutes14:47
opendevreviewMerged openstack/nova-specs master: Update the PCI in placement spec  https://review.opendev.org/c/openstack/nova-specs/+/85521815:00
bauzassean-k-mooney: your audio becomes a bit robotic15:12
stephenfinsean-k-mooney: Y̶o̶u̶'̶r̶e̶ ̶a̶ ̶w̶i̶z̶a̶r̶d̶,̶ ̶H̶a̶r̶r̶y̶!̶ You're a robot, Sean! 🤖15:15
sean-k-mooneystill ok i might have to try treaking something15:17
*** dasm is now known as dasm|off22:19

Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!