| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2021-03-01 | |||
| 18:30:27 | atmark | alias = {"vendor_id":"1344", "product_id":"51b2", "device_type":"type-PCI", "name":"nvme"} | |
| 18:32:14 | sean-k-mooney | atmark: technially that is not supported. or rather nova pci passtough is only supported when passing through stateless devices | |
| 18:32:20 | sean-k-mooney | NVME ssd are state full | |
| 18:32:32 | sean-k-mooney | its technically a security risk to sue them with nvme devices | |
| 18:32:48 | sean-k-mooney | since we will not ease the data when the vm is deleted | |
| 18:33:32 | sean-k-mooney | atmark: with that said that whitelist and alias should work | |
| 18:34:16 | sean-k-mooney | can you explain what you mean by seperate the two in teh whitelist | |
| 18:34:42 | sean-k-mooney | are you trying to tag the devices in some way and match the alias to a specific device on the host via that tag | |
| 18:36:04 | sean-k-mooney | there is no supported way to do that im affraid | |
| 18:36:18 | sean-k-mooney | the schema for the alias is https://github.com/openstack/nova/blob/master/nova/pci/request.py#L64-L97 | |
| 18:36:35 | sean-k-mooney | it does not have a way to match on the extra tags you can add in the pci whitelist | |
| 18:36:51 | sean-k-mooney | nor can it match on a pci address or similar and that is by design | |
| 18:37:40 | sean-k-mooney | the alias is not allowed to be host specific | |
| 18:37:52 | atmark | Yup. I want to tag devices that match alias to a specific device on a host. I have these NVMes which has same product and vendor ID | |
| 18:38:00 | atmark | IOMMU Group 30 41:00.0 Non-Volatile memory controller [0108]: Micron Technology Inc Device [1344:51b2] (rev 02) | |
| 18:38:01 | atmark | IOMMU Group 30 42:00.0 Non-Volatile memory controller [0108]: Micron Technology Inc Device [1344:51b2] (rev 02) | |
| 18:38:31 | sean-k-mooney | atmark: ya so we do not supprot that. i can see aded support for matching on the extra tags in teh pci whitelist | |
| 18:38:40 | sean-k-mooney | but not specificly based on address | |
| 18:39:02 | sean-k-mooney | atmark: the alais has to be the same on all compute nodes and the contolers so we need a layer of indriection | |
| 18:39:37 | sean-k-mooney | this would be a new feature unfortunetly and it still would not adress the fact that stateful devices are unsupported both upstream and downstream | |
| 18:40:08 | sean-k-mooney | stateful devices can work but there is no tenant isolation | |
| 19:00:09 | atmark | Thanks for the info. I will consider other solution then. | |
| 19:44:07 | openstackgerrit | Merged openstack/nova master: tests: Add 'nova.virt.libvirt.utils.get_arch' stub to fixture https://review.opendev.org/c/openstack/nova/+/774904 | |
| 19:44:41 | openstackgerrit | Merged openstack/nova master: tests: Remove unnecessary mock https://review.opendev.org/c/openstack/nova/+/774905 | |
| 20:48:23 | openstackgerrit | sean mooney proposed openstack/nova master: extend pci dev_type for vdpa https://review.opendev.org/c/openstack/nova/+/777481 | |
| 20:48:23 | openstackgerrit | sean mooney proposed openstack/nova master: [WIP] add vdpa nodedev parsing and interface config gen https://review.opendev.org/c/openstack/nova/+/770532 | |
| 20:48:24 | openstackgerrit | sean mooney proposed openstack/nova master: [WIP] add vdpa device detection https://review.opendev.org/c/openstack/nova/+/770533 | |
| 21:29:22 | openstackgerrit | Merged openstack/nova master: scheduler: Add 'socket' PCI NUMA affinity policy trait https://review.opendev.org/c/openstack/nova/+/772749 | |
| 21:30:36 | openstackgerrit | Sylvain Bauza proposed openstack/nova master: Bump the Compute RPC API to version 6.0 https://review.opendev.org/c/openstack/nova/+/761452 | |
| 21:32:58 | bauzas | dansmith: zuul gave me a -1 but I think i fixed the latest issues ^ | |
| 21:34:03 | bauzas | just a concern about a grenade multihost job giving me a -1 but when looking at the logs, it's about a timeout | |
| 22:11:05 | openstackgerrit | Lee Yarwood proposed openstack/nova master: nova-manage: Add libvirt update_machine_type command https://review.opendev.org/c/openstack/nova/+/774896 | |
| 22:11:06 | openstackgerrit | Lee Yarwood proposed openstack/nova master: nova-status: Add hw_machine_type check for libvirt instances https://review.opendev.org/c/openstack/nova/+/770643 | |
| 22:11:06 | openstackgerrit | Lee Yarwood proposed openstack/nova master: nova-manage: Add libvirt list_unset_machine_type command https://review.opendev.org/c/openstack/nova/+/774897 | |
| 22:11:07 | openstackgerrit | Lee Yarwood proposed openstack/nova master: libvirt: Add a config update workflow test for [libvirt]hw_machine_type https://review.opendev.org/c/openstack/nova/+/774898 | |
| 22:11:09 | openstackgerrit | Lee Yarwood proposed openstack/nova master: docs: Add admin docs for configuring and updating machine types https://review.opendev.org/c/openstack/nova/+/774899 | |
| 22:42:20 | lyarwood | sean-k-mooney: https://zuul.opendev.org/t/openstack/build/ba023003dd21428b95f27a200437dfd8/log/job-output.txt / https://review.opendev.org/c/openstack/os-vif/+/765419/6 ; had a quick look at the os-vif backports, did you plan on fixing the LC job? | |
| 22:42:44 | lyarwood | sean-k-mooney: if not I could take a look after M3 | |
| #openstack-nova - 2021-03-02 | |||
| 00:53:48 | openstackgerrit | Merged openstack/nova master: libvirt: Define and emit DeviceRemovedEvent and DeviceRemovalFailedEvent https://review.opendev.org/c/openstack/nova/+/749929 | |
| 02:03:04 | openstackgerrit | Wenping Song proposed openstack/nova master: Nova supports password encrypted VNC https://review.opendev.org/c/openstack/nova/+/622336 | |
| 02:06:42 | openstackgerrit | Merged openstack/nova master: libvirt: Remove dead code https://review.opendev.org/c/openstack/nova/+/772928 | |
| 02:17:29 | openstackgerrit | Merged openstack/nova master: rpc: Rework 'get_notifier', 'wrap_exception' https://review.opendev.org/c/openstack/nova/+/741663 | |
| 07:25:46 | openstackgerrit | Yongli He proposed openstack/nova master: Smartnic support - cyborg drive https://review.opendev.org/c/openstack/nova/+/771362 | |
| 07:25:47 | openstackgerrit | Yongli He proposed openstack/nova master: smartnic support - new vnic type https://review.opendev.org/c/openstack/nova/+/771363 | |
| 07:25:48 | openstackgerrit | Yongli He proposed openstack/nova master: smartnic support https://review.opendev.org/c/openstack/nova/+/758944 | |
| 07:40:56 | openstackgerrit | Yongli He proposed openstack/nova master: smartnic support https://review.opendev.org/c/openstack/nova/+/758944 | |
| 08:03:21 | openstackgerrit | Yongli He proposed openstack/nova master: Smartnic support - cyborg drive https://review.opendev.org/c/openstack/nova/+/771362 | |
| 08:03:23 | openstackgerrit | Yongli He proposed openstack/nova master: smartnic support - new vnic type https://review.opendev.org/c/openstack/nova/+/771363 | |
| 08:03:25 | openstackgerrit | Yongli He proposed openstack/nova master: smartnic support https://review.opendev.org/c/openstack/nova/+/758944 | |
| 08:08:15 | yonglihe | gibi: alex_xu: i still own alex_xu 2 more check in code, everything else done. | |
| 08:25:47 | gibi | sean-k-mooney: sure I will do a doodle poll similar to previous PTGs | |
| 08:25:50 | gibi | yonglihe: ack | |
| 09:10:54 | bauzas | can someone know how I could reproduce locally the tox lower-constraints job ? | |
| 09:11:00 | bauzas | context : https://storage.gra.cloud.ovh.net/v1/AUTH_dcaab5e32b234d56b626f72581e3644c/zuul_opendev_logs_f69/761452/8/check/openstack-tox-lower-constraints/f699784/testr_results.html | |
| 09:15:10 | bauzas | mmmm, there is a l-c tox target with not using use_develop | |
| 09:16:02 | bauzas | and also, I don't have py36 running | |
| 09:16:15 | bauzas | (as default) | |
| 09:17:23 | kashyap | lyarwood: Morning, when you get a min: Nova uses in-QEMU RBD driver with raw or QCOW2 format or both? | |
| 09:30:53 | stephenfin | bauzas: care to bump this +1 to +2? https://review.opendev.org/c/openstack/nova/+/775415/ | |
| 09:31:56 | hemanth_n | sean-k-mooney: can you review the PCI stat bug on stable/rocky when you get some time https://review.opendev.org/c/openstack/nova/+/761824 | |
| 09:40:47 | bauzas | stephenfin: I'm focusing on fixing the RPC API bump issues in the gate, but I can try to take a look on it later today | |
| 09:48:04 | lyarwood | kashyap: RAW only, we block qcow2 iirc | |
| 09:50:34 | kashyap | lyarwood: A QEMU dev was asking about it; do you know the reason why we block QCOW2? | |
| 09:51:08 | lyarwood | kashyap: we block it when cloning rbd volumes as rbd already does the COW for us | |
| 09:51:11 | lyarwood | kashyap: https://github.com/openstack/nova/blob/master/nova/virt/libvirt/imagebackend.py#L1038-L1041 | |
| 09:51:38 | lyarwood | kashyap: ^ that's for ephemeral storage nova controls, I'm not sure about cinder tbh | |
| 09:52:22 | lpetrut | hi: what should we do about the windows rbd patch? https://review.opendev.org/c/openstack/nova/+/763550 lyarwood would like to bump the lower constraints, sean-k-mooney was asking us not to do that :) | |
| 09:53:03 | lyarwood | kashyap: https://github.com/openstack/cinder/blob/85975fb2b63866ba6e216e621b84d571b9dbe90e/cinder/volume/drivers/rbd.py#L1525-L1530 - looks like the same logic in cinder | |
| 09:53:33 | lyarwood | lpetrut: yeah sorry about this, I don;t think sean-k-mooney got that the current os-brick version would fail with your change, at least I'm hoping that's the case. | |
| 09:53:41 | kashyap | lyarwood: /me clicks | |
| 09:54:39 | lyarwood | gibi / stephenfin / gibi ; ^ any thoughts on lpetrut's issue? The change will not work with the current os-brick version listed in our lc/requirements.txt so we have to bump right? | |
| 09:54:51 | kashyap | lyarwood: Thanks! So the QEMU storage dev was wondering if Nova doesn't use qcow2-over-RBD due to this 2019 bug: https://bugzilla.redhat.com/show_bug.cgi?id=1744525 | |
| 09:54:52 | openstack | bugzilla.redhat.com bug 1744525 in qemu-kvm "Writing data to the qcow2 image over RBD is too slow" [Medium,Assigned] - Assigned to sgarzare | |
| 09:55:07 | kashyap | lyarwood: But I don't think that's the case; as the logic in Cinder and Nova predates that bug | |
| 09:56:21 | stephenfin | lyarwood: lpetrut: it sounds like we're avoiding bumping lower-constraints because it causes a mess? | |
| 09:56:34 | lpetrut | stephenfin yep :) | |
| 09:57:05 | stephenfin | Okay, in that case I suggest we spend a small amount of time trying to resolve the damage, and drop l-c if we can't do it easily | |
| 09:57:28 | stephenfin | Everyone else has dropped them. No point in us suffering for little to no benefit | |
| 09:57:58 | lpetrut | stephenfin: I had a patch set that sync-ed nova's lower constraints with the os-brick ones but some people were concerned by the amount of changes: https://review.opendev.org/c/openstack/nova/+/763550/12..14/lower-constraints.txt | |
| 09:58:34 | lyarwood | stephenfin: have people dropped them on master? | |
| 09:58:38 | stephenfin | sean-k-mooney: fwiw, dropping indirect dependencies from l-c can cause dependency resolution to devolve into a multi-hour slog, since the combinatorial matrix of possible versions for those indirect dependencies is huuuuuuge | |
| 09:58:39 | lyarwood | stephenfin: I noticed the stable stuff | |
| 09:58:50 | stephenfin | lyarwood: they're totally gone from oslo and neutron. Likely many other projects also | |
| 10:01:05 | lyarwood | well well well | |
| 10:01:30 | lpetrut | Cinder still uses lower constraints. most of them have been bumped here: https://github.com/openstack/cinder/commit/d3ffa90baa959530eaa1cd1d4e3800fbe9148806#diff-f868e67d7bc10a25bc6baaea42ed5c763b42174505e4441349a52cf60dc007b0 | |
| 10:01:34 | lyarwood | it doesn't really resolve our issue however | |
| 10:01:44 | lyarwood | https://review.opendev.org/c/openstack/nova/+/763550/12..14/requirements.txt <- as os-brick causes this as well | |
| 10:01:47 | lyarwood | that IMHO we can't avoid | |
| 10:02:05 | lyarwood | why don't I spend some time later today breaking that out into another change you can rebase on lpetrut | |
| 10:02:25 | lyarwood | there's a load of bugfixes in there that we need anyway outside of the new Windows RBD stuff | |
| 10:02:28 | stephenfin | that's...downgrading most things? | |
| 10:02:48 | lyarwood | yeah what the | |
| 10:03:13 | stephenfin | That seems off. We won't be allowed to specify a lower limit that os-brick, but we should be able to specify a higher one | |
| 10:03:13 | lpetrut | not quite, it's flipped :) | |
| 10:03:37 | lyarwood | oh right because you reverted it so the diff is the wrong way around | |
| 10:03:39 | stephenfin | ahh | |
| 10:03:41 | stephenfin | okay :) | |
| 10:03:48 | stephenfin | phew | |
| 10:03:50 | lpetrut | https://review.opendev.org/c/openstack/nova/+/763550/12/lower-constraints.txt | |
| 10:04:00 | stephenfin | yeah, I have no issues with bumping l-c | |