| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2022-02-25 | |||
| 13:20:56 | gibi | yes, the last two days was really review heavy | |
| 13:21:05 | gibi | but I think we have a farly good FF period | |
| 13:21:10 | gibi | *fairly | |
| 13:44:27 | chateaulav | sean-k-mooney: with all the changes merged yesterday. is it best to rebase and then continue the ci job patchset | |
| 13:48:33 | kashyap | chateaulav: Thank you for the work on https://blueprints.launchpad.net/nova/+spec/pick-guest-arch-based-on-host-arch-in-libvirt-driver | |
| 13:49:05 | kashyap | chateaulav: Hehe, you also added RISC-V too | |
| 13:49:54 | kashyap | Sorry that I couldn't take a deeper look earlier! I trust the existing reviews :) | |
| 13:50:40 | chateaulav | kashyap: lol, well tried, that will be another cycle. hit some roadblocks with ovo and there is some other things in its early stages that will help for eaiser implementation later | |
| 13:50:54 | chateaulav | thanks and no worries | |
| 13:51:18 | kashyap | chateaulav: ;-) For RISC-V stuff, check out the blog of Rich Jones (of libguestfs) here: https://rwmj.wordpress.com/tag/risc-v/ | |
| 13:51:32 | kashyap | He does a lot of testing w/ new hardware. And also maintains the Fedora RISC-V | |
| 13:53:51 | chateaulav | will do, i got to the point of having it somewhat function, i was working through the loading custom firmware descriptor files for riscv becuase it uses specific uboot and opensbi | |
| 13:54:14 | chateaulav | have a lot of links and hadnt stumbled upon that one, so def will help | |
| 13:58:54 | opendevreview | Dmitrii Shcherbakov proposed openstack/nova master: Fix migration with remote-managed ports & add FT https://review.opendev.org/c/openstack/nova/+/829974 | |
| 14:02:14 | dmitriis | sean-k-mooney, gibi: addressed comments in https://review.opendev.org/c/openstack/nova/+/829974 Tried testing revert_resize with cold migration but connection info doesn't seem to be updated to the right state (of the node an instance was migrated from) - would prefer to figure out why that is in a follow-up if possible | |
| 14:02:14 | sean-k-mooney | chateaulav: am you could rebase the zuul patch altohg if you do while you are working on it its proably better to disable the other jobs | |
| 14:02:32 | sean-k-mooney | chateaulav: just comment them out until you get your jobs working | |
| 14:02:39 | sean-k-mooney | then you can uncomment them | |
| 14:02:44 | sean-k-mooney | chateaulav: it just saves ci time | |
| 14:02:55 | chateaulav | k | |
| 14:03:32 | sean-k-mooney | dmitriis: ack sure we can likely cover revert seperatly in a differnt patch | |
| 14:03:44 | sean-k-mooney | dmitriis: have you tried it locally | |
| 14:03:50 | gibi | dmitriis: thanks. I will try to look at the patch | |
| 14:04:15 | sean-k-mooney | dmitriis: as in does it work in reality and this is a func test issue or is revert actully broken | |
| 14:05:04 | dmitriis | sean-k-mooney: I've only tried in a func test form, need to do that with updated packages on a real env and 2 nodes | |
| 14:05:26 | dmitriis | still need to set up a virtualized setup with only 1 DPU - will work on that | |
| 14:10:34 | dmitriis | sean-k-mooney: fwiw, I think the issue might not be specific to remote-managed ports as I get something like this: | |
| 14:10:43 | dmitriis | KeyError: 'pci_0000_80_00_4' | |
| 14:10:43 | dmitriis | n | |
| 14:10:43 | dmitriis | -> pci_info_from_connection = self._connection.pci_info.devices[ | |
| 14:10:43 | dmitriis | > /home/ubuntu/src/opendev/nova/nova/tests/fixtures/libvirt.py(1226)verify_hostdevs_interface_are_vfs() | |
| 14:11:04 | dmitriis | I'll try adding a test case to regular SR-IOV suite and see if it fails in the same way | |
| 14:11:04 | sean-k-mooney | dmitriis: technially resize can be doen on a singel host | |
| 14:11:10 | sean-k-mooney | so you can resize to a new flavor | |
| 14:11:13 | sean-k-mooney | then tell it to revert | |
| 14:11:30 | sean-k-mooney | resize and cold migration use the same code | |
| 14:11:42 | sean-k-mooney | cold migration is implemetned as a resize to the same flaovr | |
| 14:12:30 | sean-k-mooney | dmitriis: ack, im not sure if we have revert resize test for sriov | |
| 14:12:41 | sean-k-mooney | so ya it might just be a defect in the fixture | |
| 14:12:50 | dmitriis | sean-k-mooney: yeah, I'll try to add one but I'll probably run into the same | |
| 14:29:54 | whoami-rajat | sean-k-mooney, yes, and if i would've got the feedback that the patch is not acceptable in novaclient, I would've worked on the OSC alternative before feature freeze as my patch was proposed on Jan 31st | |
| 14:30:21 | whoami-rajat | the problem is the nova side changes have merged and to make the full feature working, we need nova side changes as well | |
| 14:30:47 | whoami-rajat | even if we get the nova API patch merged, without the client, i don't see much use of it for an end user | |
| 14:31:02 | whoami-rajat | I suppose they rely on clients rather than directly hitting the API | |
| 14:32:05 | whoami-rajat | s/nova/cinder side changes have merged for the feature | |
| 14:33:47 | sean-k-mooney | well i guess we can try and do a early osc release after rc1 | |
| 14:34:16 | sean-k-mooney | technially it would be teh Zed version but that would be an option | |
| 14:34:53 | sean-k-mooney | once rc1 is tagged the stable branches get created and master reopens for the next release | |
| 14:34:57 | sean-k-mooney | thats in about 2 weeks time | |
| 14:35:38 | sean-k-mooney | so if the osc team were open to it you coudl still proably get it in a osc release before the zed release is widely available | |
| 14:36:32 | whoami-rajat | sean-k-mooney, ack, i will explore that option, thanks | |
| 14:38:59 | sean-k-mooney | it looks like we have not actully compelte ll the work items form the last ptg | |
| 14:39:03 | sean-k-mooney | whoami-rajat: https://etherpad.opendev.org/p/nova-yoga-ptg#L513 | |
| 14:39:30 | sean-k-mooney | specificaly we have not update dthe sepc template to call out the requriemtn for sdk/osc | |
| 14:39:52 | sean-k-mooney | ill try and update the zed one to make that more clear to other going forward | |
| 14:40:48 | whoami-rajat | ack, that will be helpful | |
| 15:01:11 | bauzas | ok, fwiw, starting to use the -2 hammer | |
| 15:08:05 | bauzas | gibi: I guess there is no procedural markup for the any-traits series ? | |
| 15:08:49 | gibi | bauzas: what do you mean by procedural markup? | |
| 15:08:51 | bauzas | oh shit, nevermind, found the story | |
| 15:09:06 | bauzas | gibi: I guess we can close https://storyboard.openstack.org/#!/story/2005345 ? | |
| 15:09:26 | gibi | yeeah | |
| 15:09:33 | bauzas | ok, doing it | |
| 15:09:42 | gibi | I thought my commit messages updated the story but it didn't | |
| 15:09:43 | gibi | :/ | |
| 15:09:53 | gibi | I will add a link to the story to point to the implementation topic | |
| 15:10:07 | gibi | hm | |
| 15:10:14 | gibi | the comments are there | |
| 15:10:20 | gibi | just no tasks were created | |
| 15:10:46 | gibi | ahh comments is only from the spec patch | |
| 15:10:55 | gibi | storyboard :/ | |
| 15:11:20 | bauzas | yeah :( | |
| 15:17:50 | dansmith | bauzas: are you starting the -2 hammer for features? | |
| 15:18:00 | bauzas | dansmith: I'm on it | |
| 15:18:07 | bauzas | just did the implemented scrubbing | |
| 15:18:17 | bauzas | now, next is the procedural hammer | |
| 15:18:25 | dansmith | okay | |
| 15:18:35 | dansmith | I guess that volume-backed-rebuild one is going to be off the table | |
| 15:18:47 | dansmith | I think it's missing osc support and tempest testing | |
| 15:18:51 | dansmith | bummer :( | |
| 15:19:17 | sean-k-mooney | osc support missed yes but is the api change merged? | |
| 15:19:30 | opendevreview | Jonathan Race proposed openstack/nova master: zuul-job for Adds Pick guest CPU architecture based on host arch in libvirt driver support https://review.opendev.org/c/openstack/nova/+/828372 | |
| 15:19:31 | dansmith | bauzas: the splitting helped uncover that it was introducing a data loss bug for the old path, so I'm super glad we did that | |
| 15:19:35 | dansmith | sean-k-mooney: no | |
| 15:19:43 | bauzas | dansmith: hah, good to know | |
| 15:20:07 | sean-k-mooney | dansmith: ok then ya i guess that will miss | |
| 15:20:30 | sean-k-mooney | the cinder side is done? | |
| 15:23:29 | opendevreview | Jonathan Race proposed openstack/nova master: zuul-job for Adds Pick guest CPU architecture based on host arch in libvirt driver support https://review.opendev.org/c/openstack/nova/+/828372 | |
| 15:24:49 | whoami-rajat | sean-k-mooney, yes, cinder side changes are merged | |
| 15:25:06 | whoami-rajat | bauzas, dansmith is there a chance we can get FFE for it? | |
| 15:25:30 | dansmith | whoami-rajat: up to bauzas but without osc and tempest yet it seems like a stretch :/ | |
| 15:26:01 | opendevreview | Jonathan Race proposed openstack/nova master: zuul-job for Adds Pick guest CPU architecture based on host arch in libvirt driver support https://review.opendev.org/c/openstack/nova/+/828372 | |
| 15:26:12 | whoami-rajat | i think we already discussed the OSC thing above, tempest I will complete by monday for sure | |
| 15:26:33 | bauzas | whoami-rajat: I haven't planned any FFE yet | |
| 15:27:03 | dansmith | well, I'll review the latest set in a bit regardless | |
| 15:27:26 | bauzas | whoami-rajat: I can left dansmith to reviewing the series | |
| 15:27:40 | bauzas | whoami-rajat: but my concern is about the API microversion, it's a bit late | |
| 15:28:00 | dansmith | whoami-rajat: I really want to see tempest before I even +2, FWIW | |
| 15:28:08 | dansmith | for something like this that'd always be my plan | |
| 15:29:15 | whoami-rajat | ack, I'm still working on passing the reimage_boot_volume parameter from API->Compute, bumping up RPC versions for conductor, compute, tests for it etc that is more work than i anticipated | |