| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2019-09-30 | |||
| 18:02:54 | gmann | but cell-down-support will be true for > 2.69 - https://github.com/openstack/nova/blob/961c2945491ebcea3cf1cb175a06d057155aa5a5/nova/api/openstack/compute/servers.py#L329 | |
| 18:03:24 | gmann | then it will include the instances form down cell also | |
| 18:03:29 | mriedem | no | |
| 18:03:31 | mriedem | that's not what that means | |
| 18:03:39 | mriedem | you need to look at the InstanceLister code | |
| 18:03:45 | aspiers | sean-k-mooney: I've replied and emailed the SEV gurus to ask them | |
| 18:03:53 | aspiers | sean-k-mooney: I would prefer to deal with that in a follow-up though | |
| 18:04:14 | aspiers | especially considering we don't actually know for sure yet it's an issue | |
| 18:04:32 | sean-k-mooney | yep we could | |
| 18:04:56 | sean-k-mooney | the code as written should be correct and adress the current bug | |
| 18:05:09 | mriedem | gmann: in here https://github.com/openstack/nova/blob/961c2945491ebcea3cf1cb175a06d057155aa5a5/nova/compute/multi_cell_list.py#L414 | |
| 18:05:17 | sean-k-mooney | but i think we should file a bug for the qemu-channel too | |
| 18:05:42 | sean-k-mooney | aspiers: if ye file a bug for it im +1 i just dont want to loose track of it | |
| 18:05:57 | sean-k-mooney | aspiers: we can always close it as invalid if its not required | |
| 18:06:17 | sean-k-mooney | aspiers: did ye try to boot a vm with a spice console and sev? | |
| 18:06:19 | aspiers | sean-k-mooney: I'll wait until the gurus respond, they're normally very fast and that can potentially save me the effort of filing | |
| 18:06:26 | aspiers | sean-k-mooney: yes I booted with spice fine | |
| 18:06:37 | aspiers | or at least I'm pretty sure | |
| 18:06:46 | aspiers | I tried a lot of different combos, can't remember all the details | |
| 18:07:07 | aspiers | definitely saw one minor virtio device without iommu | |
| 18:07:10 | aspiers | and it didn't crash | |
| 18:07:39 | aspiers | in fact I think I checked about that with our guru and he said don't worry about it... think it was mouse or USB input of some kind | |
| 18:09:17 | aspiers | heading off now o/ | |
| 18:09:26 | sean-k-mooney | aspiers: its used for spices usb fowradign support that allows usb devices to be hotpluged at runtime | |
| 18:09:41 | sean-k-mooney | without requireing libvirt to modify the xml | |
| 18:09:57 | sean-k-mooney | i belive it also allows spcie to do remote hotpug | |
| 18:10:22 | sean-k-mooney | e.g. a local usb to a remove vm like RDP does with drives and printers | |
| 18:15:50 | gmann | mriedem: thanks. that param in multi cell control only exception or skip to prepare the minimum response in APi view. | |
| 18:18:44 | gmann | we need to mock the same in sample tests also. | |
| 18:21:28 | mriedem | gmann_afk: ok. there are samples for 2.69 which use a down cell fixture so maybe something is just missing from that, idk. | |
| 18:56:41 | mriedem | dansmith: maybe we should finally fix this :) https://bugs.launchpad.net/nova/+bug/1781286 | |
| 18:56:41 | openstack | Launchpad bug 1781286 in OpenStack Compute (nova) "CantStartEngineError in cell conductor during reschedule - get_host_availability_zone up-call" [Medium,Triaged] | |
| 18:57:03 | mriedem | i think it's just a matter of shoving the AZ into the Selection object that we pass around in conductor | |
| 18:58:05 | mriedem | tricky thing is functional testing since we don't have a great way to poison conductor on a reschedule to not hit the API DB | |
| 18:58:39 | dansmith | ack | |
| 18:59:01 | dansmith | mriedem: in your reply you said cern is running without a superconductor.. you just mean their cell conductors have api creds right? | |
| 18:59:13 | mriedem | yeah | |
| 18:59:16 | mriedem | but, | |
| 18:59:28 | mriedem | i don't think they make a distinction between a top and cell level conductor | |
| 18:59:30 | dansmith | because unless I'm missing something, you can't not have superconductors, as you need something listening on the conductor topic that isn't going to be confused for cell traffic | |
| 19:00:18 | dansmith | I guess you could have your apis set to penalize one cell's conductors to do the super work, if that made sense in your topology (i.e. one big cell and a bunch of smaller ones or something) | |
| 19:00:19 | mriedem | hmm, i just assumed they were doing whatever devstack does for singleconductor | |
| 19:00:25 | mriedem | which still has cell0 and cell1 | |
| 19:00:35 | mriedem | right, that's what i'm assuming cern does | |
| 19:00:43 | dansmith | well, if you do, then cell1 ends up doing all the work for cell2 for example | |
| 19:00:45 | mriedem | or just a handful of big conductor | |
| 19:00:57 | mriedem | we could ask them but no one in channel | |
| 19:01:21 | dansmith | for a deployment tool, they'd have to have some way of letting you pick which cell should be burdened with the extra work | |
| 19:01:25 | dansmith | which seems less than ideal to me | |
| 19:03:15 | mriedem | as for the console proxy auth stuff, | |
| 19:03:15 | mriedem | https://specs.openstack.org/openstack/nova-specs/specs/rocky/implemented/convert-consoles-to-objects.html | |
| 19:03:25 | mriedem | "instead we will resolve the cell database issue by running console proxies per cell instead of global to a deployment, such that the cell database is local to the console proxy" | |
| 19:03:30 | mriedem | i couldn't find that in our docs | |
| 19:03:35 | mriedem | which seems...bad | |
| 19:03:46 | mriedem | like it should be in https://docs.openstack.org/nova/latest/user/cellsv2-layout.html | |
| 19:03:47 | mriedem | melwitt: ^? | |
| 19:03:55 | mriedem | did the console proxy per cell thing get into the docs? | |
| 19:04:19 | dansmith | yeah, I couldn't either, which is why I punted to her | |
| 19:04:39 | mriedem | nothing here either https://docs.openstack.org/nova/latest/admin/remote-console-access.html | |
| 19:11:25 | mriedem | mgoddard: when you say kolla-ansible is adding support for cells v2, you mean multiple cells specifically right? because cells v2 (cell0, cell1, mappings, etc) has been required since ocata | |
| 19:12:16 | dansmith | mriedem: I think he said specifically adding a cell to an existing deployment | |
| 19:13:26 | mriedem | ah yeah at the end, "I'm trying to design it such that it's possible to deploy a new cell in an existing environment" | |
| 19:13:37 | dansmith | aye | |
| 19:18:38 | mriedem | here is a docs bug for the console proxy per cell thing anyway https://bugs.launchpad.net/nova/+bug/1846045 | |
| 19:18:38 | openstack | Launchpad bug 1846045 in OpenStack Compute (nova) "Docs don't mention running console proxies per cell" [Undecided,New] | |
| 19:38:09 | openstackgerrit | Eric Fried proposed openstack/nova master: Allow versioned discovery unauthenticated https://review.opendev.org/685181 | |
| 20:14:25 | openstackgerrit | Eric Fried proposed openstack/nova master: DNM: Alt: Extract some helper functions from DriverVolumeBlockDevice https://review.opendev.org/685782 | |
| 21:03:55 | openstackgerrit | Merged openstack/nova master: make virtual pmem feature compatible with python3 https://review.opendev.org/685593 | |
| 21:04:28 | efried | dansmith: ^ | |
| 21:39:29 | markguz_ | Hi, Anyone know of way to group pci passthrough cards together based on a label or tagging? | |
| 21:40:08 | markguz_ | for example. I have a compute node with 4 Identical Fiber Channel cards. 2 are connected to SAN 1 and 2 are connected to SAN 2. | |
| 21:40:44 | markguz_ | i wanted to be able to group the cards based on what they were connected to, but so far I have not found a way of doing that | |
| 21:40:50 | markguz_ | am i out of luck? | |
| 21:52:45 | efried | markguz_: I defer to sean-k-mooney for all things pci alias, but yeah, I don't think you can e.g. have each of two aliases grouping together the cards on the same SAN. | |
| 21:55:10 | markguz_ | efried: Yeah I thought maybe I could make an alias for a pci slot, but alias doesn't support it | |
| 21:55:37 | efried | nope, only whitelist has wildcarding I think | |
| 21:56:01 | efried | once we have placement tracking of devices, your SAN is a trait... | |
| 21:56:46 | markguz_ | efried: and I suppose there's anyway to reference the whitelist directly in a flavor? | |
| 21:57:08 | efried | no, the alias is the level of abstraction available to flavors. | |
| 21:59:06 | openstackgerrit | Matt Riedemann proposed openstack/nova master: WIP: Add Selection.availability_zone field https://review.opendev.org/685807 | |
| 21:59:43 | markguz_ | efried: bummer. thanks though. | |
| 22:06:30 | openstackgerrit | Eric Fried proposed openstack/nova master: Allow versioned discovery unauthenticated https://review.opendev.org/685181 | |
| 22:39:52 | efried | mriedem: we used to have something in nova-specs that enforced conformance to the template, no?? | |
| 22:41:55 | efried | ah, it's hidden in pep8 | |
| 22:59:12 | openstackgerrit | Eric Fried proposed openstack/nova-specs master: Add 'Core Liaison' section to spec template https://review.opendev.org/685857 | |
| #openstack-nova - 2019-10-01 | |||
| 01:33:39 | openstackgerrit | Eric Fried proposed openstack/nova master: Allow versioned discovery unauthenticated https://review.opendev.org/685181 | |
| 02:47:08 | SonPham | hi | |
| 02:47:19 | SonPham | had any project to create live-snapshot for openstack? | |
| 02:47:48 | SonPham | snapshot with save ram state | |
| 03:01:28 | brinzhang_ | Now, nova not support to do for that | |
| 03:04:55 | openstackgerrit | Seyeong Kim proposed openstack/nova master: block_device: Optionally recreate attachments when refreshing connection_info https://review.opendev.org/579004 | |
| 03:07:43 | brinzhang_ | SonPham: sorry for above said, current, cannot do a live-snapshot for hte paused server of libvirt/qemu driver. | |
| 03:08:13 | brinzhang_ | paused and shutdown status server. | |
| 03:10:13 | melwitt | dansmith, mriedem: the doc reference got removed here https://github.com/openstack/nova/commit/009fd0f35bcb88acc80f12e69d5fb72c0ee5391f | |
| 03:26:16 | SonPham | brinzhang_ Can we save VM state (save Ram) in the snapshot? | |
| 03:31:27 | brinzhang_ | SonPham: I think you can, bug I am not sure, you can review that and try, with "disable_libvirt_livesnapshot" configuration in you env. | |
| 03:31:43 | brinzhang_ | SonPham: https://docs.openstack.org/nova/stein/configuration/config.html#workarounds.disable_libvirt_livesnapshot | |
| 03:54:41 | SonPham | brinzhang_ I'm using libvirt to create snapshot for instance | |
| 08:33:17 | mgoddard | mriedem: yeah, I meant multiple cells. We would be quite behind to not have any support for cells v2... | |
| 09:57:58 | bauzas | gibi: can I ask you something about how to create a new allocation ? | |