| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2024-11-28 | |||
| 15:09:08 | sean-k-mooney | ishanwar: you can techinally do that but its not hwo developemnt is normally done. nova is a distirbute system with several external depencies | |
| 15:09:41 | sean-k-mooney | so the normall approch is to use devstack https://github.com/openstack/devstack to create a devleopment enviornment in a vam | |
| 15:09:43 | sean-k-mooney | *vm | |
| 15:18:38 | zigo | sean-k-mooney: My db is comming from another cluster, just testing db upgrades and online_data_migrations. Could that be the reason? | |
| 15:19:37 | zigo | FYI, that's form Antelope to Caracal. | |
| 15:19:42 | zigo | from | |
| 15:28:50 | sean-k-mooney | populate_instance_compute_id should be using data form within the cell db | |
| 15:29:31 | sean-k-mooney | so if your doign this on a local backup ectra then it should not cause issues that im aware of | |
| 15:30:47 | sean-k-mooney | this is what its actully doing https://github.com/openstack/nova/blob/master/nova/objects/instance.py#L1357-L1377 | |
| 15:31:47 | sean-k-mooney | for each instance we get the node and host values and lookup the compute node id then set that in teh instance compute_id column | |
| 15:32:32 | sean-k-mooney | so that shoudl be relitivly quick although if you have a lot of instnaces then it will take a while | |
| 15:33:16 | zigo | That cluster has ~8k VM, though many deleted, I guess. | |
| 15:33:42 | zigo | It's rather fast, what I'm worried about, is the None thingy, plus compute nodes that we decomitionned. | |
| 15:33:44 | sean-k-mooney | we are reading deleted isntanfe too | |
| 15:34:01 | zigo | Yeah, and I don't understand why. | |
| 15:34:25 | sean-k-mooney | its backpopulating deleted but not archived isntances | |
| 15:34:33 | sean-k-mooney | to keep it consitent | |
| 15:34:44 | zigo | Oh, maybe I should just archive them then ... | |
| 15:35:30 | sean-k-mooney | so if the compute node is delete then this will skip them | |
| 15:35:46 | sean-k-mooney | but if your rungin this in a loop it will never fully complete | |
| 15:37:12 | zigo | Hum... I do have the ::nova::cron::archive_deleted_rows thingy ... :P | |
| 15:37:14 | sean-k-mooney | in theory if you are deleting a compute node shoudl shoudl also be deleting any instance that are on it or move them to a diffent host first | |
| 15:37:41 | sean-k-mooney | but there are bugs that can lead to instanfce beign in a wereid state sometimes | |
| 16:09:38 | sean-k-mooney | bauzas: as an fyi you have not created the 2025.1 series in https://launchpad.net/nova yet or set it as the active developemnt release, it still says 2024.2 | |
| 16:09:50 | sean-k-mooney | bauzas: proably needs to be done on the other delvierables too | |
| 16:10:05 | bauzas | hmmm, no, we already have blueprints for 2025.1 | |
| 16:10:20 | bauzas | but indeed, maybe the active release is still 2024.2 | |
| 16:13:19 | bauzas | sean-k-mooney: okay, now fixed | |
| 16:13:25 | bauzas | https://launchpad.net/nova/+series | |
| 16:15:23 | sean-k-mooney | cool | |
| 16:15:45 | sean-k-mooney | im just looking at fixing watchers trackign and im using novas config as a refence | |
| 16:15:54 | sean-k-mooney | so i just noticed that when comparing | |
| 16:40:25 | opendevreview | Merged openstack/nova master: Revert "[libvirt] Live migration fails when config_drive_format=iso9660" https://review.opendev.org/c/openstack/nova/+/909122 | |
| 16:56:10 | sean-k-mooney | woot now to backport that... | |
| 16:56:30 | opendevreview | sean mooney proposed openstack/nova stable/2024.2: Revert "[libvirt] Live migration fails when config_drive_format=iso9660" https://review.opendev.org/c/openstack/nova/+/936645 | |
| 18:26:12 | opendevreview | ribaudr proposed openstack/nova master: Refactor response schemas for share API https://review.opendev.org/c/openstack/nova/+/936656 | |
| 19:23:16 | opendevreview | Balazs Gibizer proposed openstack/placement master: WIP: Test mitigations for candidate explosion https://review.opendev.org/c/openstack/placement/+/936658 | |
| 21:41:45 | opendevreview | Merged openstack/nova master: Support resuming an instance with shares (compute manager part) https://review.opendev.org/c/openstack/nova/+/860285 | |
| #openstack-nova - 2024-11-29 | |||
| 04:22:04 | opendevreview | Michael Still proposed openstack/nova master: libvirt: direct SPICE console database changes https://review.opendev.org/c/openstack/nova/+/926877 | |
| 04:22:04 | opendevreview | Michael Still proposed openstack/nova master: libvirt: direct SPICE console object changes https://review.opendev.org/c/openstack/nova/+/926876 | |
| 04:22:05 | opendevreview | Michael Still proposed openstack/nova master: libvirt: Add extra spec for sound device. https://review.opendev.org/c/openstack/nova/+/926126 | |
| 04:22:05 | opendevreview | Michael Still proposed openstack/nova master: libvirt: allow direct SPICE connections to qemu https://review.opendev.org/c/openstack/nova/+/924844 | |
| 04:22:06 | opendevreview | Michael Still proposed openstack/nova master: libvirt: Add extra specs for USB redirection. https://review.opendev.org/c/openstack/nova/+/927354 | |
| 10:35:16 | r-taketn | Hello. We are planning to contribute a new feature for building Arm-Confidential Computing Architecture(CCA)-enabled instances. Could you please review the following e-mail? (sorry, mail title is bad) https://lists.openstack.org/archives/list/openstack-discuss@lists.openstack.org/thread/244U52T3HINVWUFSMPMA45A67BUPAQGK/ I'm not sure it is acceptable for OpenStack because Arm-CCA enabled hardware is not yet available to our knowledge. | |
| 10:35:47 | r-taketn | Your feedback on the acceptance of this contribution would be greatly appreciated. | |
| 14:47:02 | opendevreview | Merged openstack/nova master: Add helper methods to rescue/unrescue shares https://review.opendev.org/c/openstack/nova/+/860286 | |
| 14:50:27 | kevko | Hi folks, trying to snapshot whole instance (which will produce 0 bytes glance image with metadata + cinder volume snapshot) ...i am using ceph, when i want to rebuild the server with openstack server rebuild --reimage-boot-volume --image ${server_snapshot} --os-compute-api-version 2.93 ${server_id} it is failing with something like this | |
| 14:50:58 | kevko | https://paste.openstack.org/show/bMAK9nICw823XtdGaFaG/ | |
| 14:51:01 | kevko | any help please ? | |
| 14:51:27 | kevko | I can see that it's trying to process 0 bytes image from glance ... | |
| 14:51:36 | kevko | what i am doing bad ? | |
| 14:51:46 | kevko | some ceph settings ? copy on write ? what :( ? | |
| 14:51:53 | kevko | can anybody help ? thank you very much | |
| 14:54:15 | sean-k-mooney | kevko: this is a boot form volume instance correct | |
| 14:56:11 | sean-k-mooney | the boot form volume image rebuild was mainly implemeted for rebuilding form a normal image not a snapshot | |
| 14:56:20 | sean-k-mooney | not both shoudl be supprotable | |
| 14:56:57 | sean-k-mooney | but im not sure how well tested this is and if it works today. | |
| 14:58:38 | sean-k-mooney | im not sure if https://review.opendev.org/c/openstack/nova/+/930754 would adress this or not | |
| 14:58:59 | kevko | sean-k-mooney: this is my script to test | |
| 14:59:01 | kevko | https://paste.openstack.org/show/bpEXTVEN6E6Jc0ZTnXta/ | |
| 14:59:59 | sean-k-mooney | ok so you takign the server create form precreated volume flow | |
| 15:00:13 | sean-k-mooney | which is one of the two boot form volume workflows | |
| 15:00:48 | sean-k-mooney | then your creating a normla snapshot and then rebuildign to that | |
| 15:00:58 | kevko | sean-k-mooney: yeah, that should work no ? | |
| 15:01:31 | kevko | it's kinda a new feature to rebuild with cinder backed volume ...from 2.93 microversion | |
| 15:01:47 | kevko | I will be very grateful if you help me :) | |
| 15:02:18 | sean-k-mooney | so the reason im hesitating is i dont know if that supprot the special optimiation that we do when cinder and glance share the same storage backend | |
| 15:04:01 | sean-k-mooney | cidner and glance need to be configured in a spefic way in general to allwo them to do thing snaphosts when creatign new voluems form an image/snapshot for example | |
| 15:04:41 | sean-k-mooney | the main testing around https://specs.openstack.org/openstack/nova-specs/specs/zed/implemented/volume-backed-server-rebuild.html i belive was done with out ceph | |
| 15:05:00 | sean-k-mooney | so glance on file/swirft and cidner on lvm | |
| 15:05:16 | kevko | sean-k-mooney: i was able to catch incoming files with rsync in infinity loop | |
| 15:05:18 | kevko | sean-k-mooney: https://ibb.co/KK2KzfR | |
| 15:05:53 | kevko | sean-k-mooney: so i was wondering if there is some bug maybe ? because there is a 0 bytes image and also bigger one (cirros) | |
| 15:06:57 | sean-k-mooney | so what the cinder backend shoudl be doing in this case is the volume snapshot shoudl be creating a snapshot in ceph | |
| 15:07:01 | kevko | sean-k-mooney: https://ibb.co/F4NJDhB | |
| 15:07:14 | sean-k-mooney | and hten the rebuidl on the cider driver side should be just askign ceph to restore it | |
| 15:07:35 | sean-k-mooney | there should be no data trasfer in either code path in this case | |
| 15:07:59 | kevko | https://ibb.co/48gFG7f << attempt to import 0 bytes image into volumes pool | |
| 15:08:31 | sean-k-mooney | yep which impleis that you have an issue in how you have configred cinder an glance to tlak to each other | |
| 15:08:57 | sean-k-mooney | you are using the same ceph cluster for glance and cinder correct? | |
| 15:09:24 | kevko | sean-k-mooney: sure, I'm not an expert ... but at least I'm trying to follow the threads and I'm getting further and further ... and it's strange that internal Ceph isn't used ... and that the data is being copied... That also seemed strange to me. | |
| 15:09:34 | kevko | sean-k-mooney: yes, same cluster ... | |
| 15:10:09 | sean-k-mooney | https://docs.openstack.org/cinder/latest/admin/volume-backed-image.html | |
| 15:10:10 | kevko | sean-k-mooney: what do you think how i configured glance<->cinder to talk each other ? | |
| 15:10:21 | sean-k-mooney | have you confiugred glance ot supprot volume backed images | |
| 15:10:53 | sean-k-mooney | kevko: this is really not a nova issue by the way the error is on the cinder side | |
| 15:12:24 | sean-k-mooney | i belvie you need to configure that for https://specs.openstack.org/openstack/cinder-specs/specs/liberty/clone-image-in-glance-cinder-backend.html to work | |
| 15:13:11 | kevko | sean-k-mooney: this is my conf https://paste.openstack.org/show/bc0ZzJIdENRaizbJR1vG/ | |
| 15:14:06 | sean-k-mooney | part of me hopes SECRET_OMMITED is the acutal password and your are just trolling peopel :P | |
| 15:14:47 | kevko | :D | |
| 15:14:56 | kevko | that's not my pass :P | |
| 15:15:00 | kevko | hahaaha | |
| 15:15:06 | sean-k-mooney | it would make me sad if it was becasue now you woudl have to change it | |
| 15:15:25 | kevko | sean-k-mooney: i know what i am doing :) | |
| 15:15:57 | kevko | only this snapshots i am debuggign 2 days and dont' know what else i can try | |
| 15:15:58 | sean-k-mooney | so im not an expert ofn this but looking at your config, glance is not configured to show "show_multiple_locations" | |
| 15:16:31 | sean-k-mooney | which mean that cidner cant detect that its the same ceph cluster | |
| 15:16:42 | sean-k-mooney | and it can jsut ask ceph to do the restore | |
| 15:16:54 | sean-k-mooney | the issue with show_multiple_locations is that it has security implciations | |
| 15:17:35 | sean-k-mooney | it makes the ceph storage url "rbd://..." visable n the responce to a normal user | |