| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2023-01-11 | |||
| 12:24:50 | kashyap | (Answering the earlier question) | |
| 12:24:53 | sean-k-mooney | kashyap: ya i didnt think we did but wantted to confirm | |
| 12:25:11 | kashyap | sean-k-mooney: I'll let some actual tests with real CPU models be done by QE to see the existing patch's impact | |
| 12:25:16 | kashyap | That way we have clear evidence. | |
| 12:25:55 | sean-k-mooney | ok i need to go take my blood pressue medication and do a few bits. brb | |
| 12:26:08 | kashyap | Sure, take care! And thanks for the discussion | |
| 12:54:51 | zigo | one colleague of mine says it's because the image was deleted first. Is he right? | |
| 12:54:51 | zigo | https://paste.opendev.org/show/bRFOvaXxQxs5vVdwzVRV/ | |
| 12:54:51 | zigo | Last night, we had a case of crash of nova when deleting a VM: | |
| 12:55:20 | sean-k-mooney | no i dont think so | |
| 12:55:39 | sean-k-mooney | this is related to the nvram not the glance/vm disk image | |
| 12:55:43 | zigo | How come Nova can't "undefine domain with nvram" then? | |
| 12:56:39 | zigo | Was this fixed in a version higher than Victoria? | |
| 12:56:41 | sean-k-mooney | it might be related to https://bugs.launchpad.net/nova/+bug/1785123 | |
| 12:57:06 | zigo | Oh, thanks. | |
| 12:57:18 | sean-k-mooney | zigo: there have been some nvram/uefi issues fixed since then yes | |
| 12:57:52 | sean-k-mooney | https://review.opendev.org/q/project:openstack/nova+nvram | |
| 12:58:29 | sean-k-mooney | https://review.opendev.org/c/openstack/nova/+/335512 that seams to be the most driectly related | |
| 12:59:29 | zigo | It's abandonned though ... :/ | |
| 12:59:35 | sean-k-mooney | https://github.com/openstack/nova/blob/master/nova/virt/libvirt/guest.py#L298 | |
| 12:59:44 | sean-k-mooney | yes but the fix was condtionaly added on master | |
| 12:59:48 | sean-k-mooney | if uefi is supported | |
| 12:59:56 | sean-k-mooney | so we fixed it in a differnt patch | |
| 13:00:11 | sean-k-mooney | https://github.com/openstack/nova/commit/539d381434ccadcdc3f5d58c2705c35558a3a065 | |
| 13:00:33 | sean-k-mooney | hum apparently in ocata | |
| 13:00:51 | sean-k-mooney | zigo: what version of qemu are you using | |
| 13:00:55 | sean-k-mooney | sory libvirt | |
| 13:01:14 | zigo | 7.0.0 | |
| 13:01:19 | zigo | (the one from Bullseye) | |
| 13:02:28 | sean-k-mooney | i wonder what support_uefi is set too | |
| 13:02:52 | opendevreview | Merged openstack/nova master: Support unshelve with PCI in placement https://review.opendev.org/c/openstack/nova/+/854616 | |
| 13:04:32 | sean-k-mooney | zigo: so we do this https://github.com/openstack/nova/blob/70fd0cfc8ee2b541ffc1f9feb129314965d1670c/nova/virt/libvirt/driver.py#L1349-L1351 | |
| 13:05:54 | zigo | Victoria has the same code... | |
| 13:06:17 | zigo | So you see... instance.image_meta.properties.get | |
| 13:06:30 | zigo | Is this taken from the *image* ?!? | |
| 13:06:58 | zigo | Or is it a copy of the image meta? | |
| 13:07:23 | sean-k-mooney | a copy but i assume its uefi | |
| 13:09:39 | sean-k-mooney | you might be higing a kernel api change where some parmater changed for 1/0 to y/n | |
| 13:09:59 | sean-k-mooney | can you check the output of virsh domcapabilities | |
| 13:11:05 | zigo | https://paste.opendev.org/show/bEp66voE4chYd3lF3eme/ | |
| 13:11:10 | zigo | What am I looking for? | |
| 13:12:14 | zigo | The <os supported='yes'> bits? | |
| 13:12:22 | sean-k-mooney | <loader supported='yes'> | |
| 13:12:31 | sean-k-mooney | line 12 | |
| 13:12:51 | zigo | Is it supposed to be 0/1 instead ? | |
| 13:13:14 | sean-k-mooney | no i was thinkin of a change for secure boot | |
| 13:13:18 | sean-k-mooney | which is seperate | |
| 13:16:12 | opendevreview | Merged openstack/nova stable/train: Adapt websocketproxy tests for SimpleHTTPServer fix https://review.opendev.org/c/openstack/nova/+/866201 | |
| 13:18:09 | zigo | sean-k-mooney: That bit of code is supposed to remove the disk containing the NVRAM values, right? | |
| 13:18:54 | sean-k-mooney | its not really a disk its a file that is uesd to store the uefi firmware data | |
| 13:19:09 | zigo | Right, that's what I had in mind. | |
| 13:19:26 | sean-k-mooney | but yes its to tell libvirt that its oke to delete the domain | |
| 13:19:47 | sean-k-mooney | inclucding undefining the nvram | |
| 13:38:59 | zigo | According to this patch's patch header https://review.opendev.org/c/openstack/nova/+/621646/ the UEFI NVRAM variable store isn't preserved on stop/start, hard reboot, cold migration, resize and live migration... | |
| 13:39:07 | zigo | Is it still the case? | |
| 13:41:17 | sean-k-mooney | that i am not sure about. i know that was a proablem in the past | |
| 13:42:17 | sean-k-mooney | i think this bug is still a thing | |
| 13:42:54 | sean-k-mooney | honestly i proably should escalte this internally too becasue this keeps getting stalled out | |
| 13:43:04 | sean-k-mooney | the people that started fixing it have moved on form openstack | |
| 14:38:03 | opendevreview | Merged openstack/nova master: Add mock to avoid loading guestfs in unit test https://review.opendev.org/c/openstack/nova/+/862769 | |
| 14:47:09 | opendevreview | Jorge San Emeterio proposed openstack/nova-specs master: Review usage of oslo-privsep library on Nova https://review.opendev.org/c/openstack/nova-specs/+/865432 | |
| 15:14:35 | dansmith | 2023-01-11 07:00:50,878 WARNING [oslo_messaging.rpc.client] Using RPCClient manually to instantiate client. Please use get_rpc_client to obtain an RPC client instance. | |
| 15:14:47 | dansmith | thousands of those in each run now.. I guess something in o.msg changed | |
| 15:24:00 | gibi | dansmith: I think https://review.opendev.org/c/openstack/requirements/+/869340 pulled in https://review.opendev.org/c/openstack/oslo.messaging/+/862419 that has the new warning | |
| 15:25:35 | opendevreview | ribaudr proposed openstack/nova-specs master: Allow local scaphandre directory to be mapped to an instance using virtiofs https://review.opendev.org/c/openstack/nova-specs/+/861881 | |
| 15:31:15 | opendevreview | ribaudr proposed openstack/nova-specs master: Allow local scaphandre directory to be mapped to an instance using virtiofs https://review.opendev.org/c/openstack/nova-specs/+/861881 | |
| 15:47:49 | gmann | dansmith: replied and updated the rbac default switch change, please check when you have time https://review.opendev.org/c/openstack/nova/+/866218 | |
| 15:48:48 | gmann | I have pushed devstack changes (depends-on) to keep running all existing jobs on old default and new jobs run with new defaults. After we do 2023.1 release we can switch it to run all existing jobs to new defaults and one job to run on old defaults | |
| 15:50:39 | dansmith | cool | |
| 17:58:35 | opendevreview | sean mooney proposed openstack/placement master: support multiple config files with apache https://review.opendev.org/c/openstack/placement/+/869863 | |
| 17:58:59 | sean-k-mooney | ^ is the placement version fo https://review.opendev.org/c/openstack/nova/+/867162 | |
| 20:12:31 | opendevreview | Dan Smith proposed openstack/nova master: Pass service ref to init_host(), if exists https://review.opendev.org/c/openstack/nova/+/863916 | |
| 20:12:31 | opendevreview | Dan Smith proposed openstack/nova master: Add virt/node module for stable uuids https://review.opendev.org/c/openstack/nova/+/863915 | |
| 20:12:32 | opendevreview | Dan Smith proposed openstack/nova master: Make resource tracker use UUIDs instead of names https://review.opendev.org/c/openstack/nova/+/863919 | |
| 20:12:32 | opendevreview | Dan Smith proposed openstack/nova master: Add get_available_node_uuids() to virt driver https://review.opendev.org/c/openstack/nova/+/863917 | |
| 20:12:33 | opendevreview | Dan Smith proposed openstack/nova master: WIP: Detect host renames and abort startup https://review.opendev.org/c/openstack/nova/+/863920 | |
| 20:12:33 | opendevreview | Dan Smith proposed openstack/nova master: Persist existing node uuids locally https://review.opendev.org/c/openstack/nova/+/863918 | |
| 20:38:11 | opendevreview | Dan Smith proposed openstack/nova master: Make resource tracker use UUIDs instead of names https://review.opendev.org/c/openstack/nova/+/863919 | |
| 20:38:12 | opendevreview | Dan Smith proposed openstack/nova master: WIP: Detect host renames and abort startup https://review.opendev.org/c/openstack/nova/+/863920 | |
| 20:38:12 | opendevreview | Dan Smith proposed openstack/nova master: Persist existing node uuids locally https://review.opendev.org/c/openstack/nova/+/863918 | |
| 21:24:04 | opendevreview | Merged openstack/nova master: Support same host resize with PCI in placement https://review.opendev.org/c/openstack/nova/+/854441 | |
| #openstack-nova - 2023-01-12 | |||
| 05:06:05 | opendevreview | Merged openstack/nova master: Test reschedule with PCI in placement https://review.opendev.org/c/openstack/nova/+/854626 | |
| 05:12:25 | opendevreview | Merged openstack/nova master: Test multi create with PCI in placement https://review.opendev.org/c/openstack/nova/+/854663 | |
| 07:26:01 | opendevreview | Tobias Urdin proposed openstack/nova master: Use get_rpc_client helper from oslo.messaging https://review.opendev.org/c/openstack/nova/+/869900 | |
| 07:26:23 | tobias-urdin | gibi: dansmith ^ maybe that | |
| 08:42:13 | auniyal | Hi sean-k-mooney, gibi | |
| 08:42:23 | auniyal | how can I validate this bug - https://bugs.launchpad.net/nova/+bug/1996732 | |
| 08:42:39 | auniyal | like after my changes | |
| 08:46:40 | gibi | stephenfin: if you have time could you get back to this https://review.opendev.org/c/openstack/nova/+/854924 this is now the bottom of the PCI series and I fixed your comments | |
| 08:46:45 | gibi | tobias-urdin: added to my queue | |
| 08:48:34 | gibi | auniyal: I think you can trigger late affinity check failures in a func test by injecting a fault there | |
| 08:50:47 | gibi | and observing the effect of the BuildFailureWeigher can be done by crafting the settings of the existing weighers so that thd BuildFailureWeigher is overrepresented in the weighing result | |
| 08:54:32 | gibi | tobias-urdin: I left feedback there | |
| 08:54:39 | gibi | thanks for proposing a fix for this | |
| 08:56:47 | tobias-urdin | gibi: ack, thx! | |
| 09:10:26 | gibi | bauzas: I've approved the privsep refactor spec | |
| 09:10:34 | bauzas | thanks | |
| 09:10:44 | bauzas | I was looking into it | |
| 09:10:50 | bauzas | but I need to go to Uggla's spec | |
| 09:11:10 | gibi | I saw your +2 from before and I had a -1 that was answered so with sean-k-mooney +2 I felt we have corum | |