Earlier  
Posted Nick Remark
#openstack-nova - 2022-07-28
10:30:22 kashyap Yes, you did; I missed to refresh; thx!
10:30:31 kashyap bauzas: Also check wallaby, please, while you have the context
10:30:45 kashyap It has gibi's ACK - https://review.opendev.org/c/openstack/nova/+/851206
10:30:55 bauzas I'll visit the wallabies
10:31:40 bauzas you sur'ly (ussuri)
10:32:01 bauzas and then I'll take the train
10:32:26 bauzas oh, but again, I missed Victoria
10:34:04 kashyap Heh
10:34:07 kashyap Thanks!
10:34:08 kashyap Yes
10:38:04 opendevreview Balazs Gibizer proposed openstack/nova master: use force down to stabilize the test https://review.opendev.org/c/openstack/nova/+/851368
10:50:35 stephenfin gibi: sean-k-mooney: I missed your ping yesterday RE: the mock stuff. That sounds like a lot of work. Damn it /o\
10:50:56 gibi stephenfin: I'm just about to start looking into it to see how much work to fix it
10:51:16 gibi so I hope I will have a better understanding by the end of today
10:51:20 stephenfin Jinx. I was about to say I can try to play around with it but I don't know if I can spend days at it, heh
10:51:36 stephenfin You saw efried's pointer to other projects that have solved this problem already?
10:51:41 gibi yes I saw
10:51:44 stephenfin sweet
10:51:52 gibi we need efried back in nova!
10:52:22 stephenfin I mean, it's not like he needs to sleep
10:52:36 stephenfin You can sleep when you're dead etc. etc. ;-)
10:53:14 gibi yepp, exactly like that :D
10:53:30 gibi I remember we had such poster in the office a couple of years back.
10:57:11 gibi bauzas, sean-k-mooney: can we land this? https://review.opendev.org/c/openstack/os-traits/+/850538 I hate to see the red CI in my series due to that
11:01:44 sean-k-mooney ah that am ill review it now in a sec just finishing an email but ya we will need to do a releas but i can propose that once its merged
11:02:36 bauzas gibi: just a +1 nit https://review.opendev.org/c/openstack/os-traits/+/850538
11:02:48 bauzas tl;dr: singular or plural ?
11:03:22 bauzas if this is for the root RP, ie. the compute host itself, I guess we could make it plural
11:03:30 gibi it will be on every PCI RP
11:03:32 bauzas if this is for a nested RP
11:03:35 bauzas then singular
11:03:35 gibi but wont be on the root
11:03:49 bauzas ok, then you answered my question, will +2
11:03:56 gibi thanks
11:04:01 sean-k-mooney i would go plural
11:04:17 bauzas sean-k-mooney: if this is a PCI RP, then singular
11:04:18 sean-k-mooney based on ACCELERATORS and REMOTE_MANAGED_PORTS
11:04:34 bauzas like "I can manage myself"
11:05:03 sean-k-mooney this is kind of like sheep
11:05:17 sean-k-mooney 1 sheep 2 sheep
11:05:27 bauzas damn english folks
11:05:29 sean-k-mooney either works but i think plural is more correct
11:05:46 sean-k-mooney well ok
11:05:49 sean-k-mooney no your right
11:05:51 sean-k-mooney singulaer
11:05:58 sean-k-mooney sicne we have 1 RP per PF
11:06:09 sean-k-mooney i was thinking about the VF inventories
11:06:12 gibi we have 1 PR per multiple VFs
11:06:13 gibi :)
11:06:21 gibi RP
11:06:23 sean-k-mooney ya so this is correct as is
11:06:27 gibi OK
11:06:43 bauzas yeah
11:06:49 bauzas I wasn't discussing about SRIOV
11:06:59 bauzas this is for a PCI ID
11:07:01 gibi btw, bauzas, this trait was not mentioned in the spec so I added a comment to the merged spec now and I will push a follow up on the spec to sync it with the implementation reality
11:07:03 bauzas hence singular
11:07:20 sean-k-mooney gibi: we can update the spec later
11:07:25 gibi yeah I
11:07:29 sean-k-mooney based on the final merged implemeantion
11:07:30 gibi I'm planning to do so
11:07:32 bauzas gibi: ok, so I wasn't lazy then :D
11:07:39 bauzas :ô
11:07:39 bauzas :ô
11:07:40 bauzas :p
11:07:42 gibi I'm collecting the comments there to know what to update
11:07:51 sean-k-mooney we also move the logic to the compute manager
11:07:55 gibi sean-k-mooney: yepp
11:08:11 sean-k-mooney i dont know if we eplxitly said tiwoudl be in the driver before but its good to document
11:08:32 bauzas gibi: I'll try to look at your series before tuesday eob
11:09:06 bauzas oh wait
11:09:09 gibi bauzas: when you look try to focus on the high level direction and architecture please
11:09:18 bauzas gibi: https://github.com/openstack/os-traits/tree/master/os_traits/hw
11:09:34 gibi we discussed with sean-k-mooney that this is not a hw trait
11:09:37 bauzas could be hw
11:09:40 gibi but more like a capability trait
11:09:43 gibi of the compute service
11:09:50 bauzas meh ok
11:12:52 opendevreview Merged openstack/os-traits master: Add COMPUTE_MANAGED_PCI_DEVICE trait https://review.opendev.org/c/openstack/os-traits/+/850538
11:12:59 gibi bauzas, sean-k-mooney: thanks!
11:13:35 gibi I'm going to push a os-traits release
11:15:39 sean-k-mooney lets wait a sec
11:15:40 bauzas damn
11:15:41 bauzas | nova               | 11            | 22           | 50.0              |
11:15:47 bauzas 50% of blind rechecks
11:15:49 bauzas man !
11:15:56 sean-k-mooney it would be nice to do https://review.opendev.org/c/openstack/os-traits/+/844336 too
11:16:38 bauzas don't people read our meeting ? :)=
11:16:40 sean-k-mooney ricolin: ^ coudl you respin that quickly and remvoe the none
11:17:23 sean-k-mooney well
11:17:31 sean-k-mooney ricolin: we could leave it i guess
11:17:45 sean-k-mooney ah actully no
11:18:08 sean-k-mooney vioummu=non shouuld not force libvirt
11:18:32 sean-k-mooney so we should remove the trait and not request a trait if you set the exptra spec to none
11:19:00 sean-k-mooney gibi: bauzas alernitivly i could adress my own nit in that patch and ye could review
11:19:09 gibi sean-k-mooney: go for it
11:19:13 gibi I can respin the releasenotes
11:19:22 gibi s/releasenotes/release/
11:21:26 opendevreview sean mooney proposed openstack/os-traits master: Add traits for vIOMMU https://review.opendev.org/c/openstack/os-traits/+/844336
11:22:30 sean-k-mooney oh the depends on is wrong

Earlier   Later