Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2019-09-04
15:57:00 sean-k-mooney by the way i was kind of expecting that might fail im just reporing what im finding as i try to break it
15:57:26 artom Totally, this sort of testing is great
15:57:51 openstackgerrit Adam Spiers proposed openstack/nova master: Enable booting of libvirt guests with AMD SEV memory encryption https://review.opendev.org/666616
15:58:07 openstackgerrit Merged openstack/nova master: Refactor MigrationTask._execute https://review.opendev.org/679651
15:58:07 artom Wait, would live migrating to an invalid pinning situation go through?
15:58:17 openstackgerrit Merged openstack/nova master: unit test: do not fill rp mapping for failed re-schedule https://review.opendev.org/679653
15:58:34 sean-k-mooney ok cold migration of a new instance works fine
15:58:47 artom I guess as long as the host CPUs are online, we don't check whether they're part of vcpu_pin_set during the LM
15:58:59 sean-k-mooney ill check reverting to master on the compute node then master only
15:59:01 artom So it'd make sense for that to fail in a subsequent operation
15:59:46 sean-k-mooney artom: right libvirt checks they are are a thing but libvirt is not aware of the vcpu_pin_set
16:00:10 sean-k-mooney and i have intentionally configred theyse hsot so the vcpu_pin_sets on each host does not overlap
16:00:44 artom Right, to make sure the old broken behaviour is happening
16:01:13 sean-k-mooney yes
16:01:25 sean-k-mooney and to give a really hard edgecase
16:02:05 aspiers kashyap: I've just built an image with an experimental kernel and grub secure boot fails due to "invalid signature"
16:02:16 aspiers kashyap: any idea how to tell it the signature is trustworthy?
16:02:17 sean-k-mooney one server is set up to only allow resouce on numa node 0 the other for node 1 only and it have the cpu_shared_set configure to use the opistie numa node to the vcpu_pin_set
16:03:07 kashyap aspiers: Yeah, "experimental kernal" is the keyword :-)
16:03:55 sean-k-mooney aspiers: it has not been sgined with a key that is in the secure boot keystore would be my guess
16:04:10 kashyap So, there is a manual way to make it "trustworthy"
16:04:27 kashyap aspiers: But I don't see a document for it. Last I checked with an admin who does it, said it's quite tedious
16:05:27 openstackgerrit Matt Riedemann proposed openstack/nova master: Add delete_on_termination to volume-attach API https://review.opendev.org/673133
16:05:35 sean-k-mooney dansmith: artom os i have a proablem with 1 node on master and the other node on artoms code
16:05:56 sean-k-mooney on startup the agent on master code is raising "ServiceTooOld: This service is older (v39) than the minimum (v40) version of the rest of the deployment."
16:06:05 donnyd Is there a way to ask nova for a specific trait on an instance without configuring a flavor first?
16:06:17 openstackgerrit Adam Spiers proposed openstack/nova master: Enable booting of libvirt guests with AMD SEV memory encryption https://review.opendev.org/666616
16:06:28 artom sean-k-mooney, update the database I think
16:06:35 kashyap aspiers: If Secure Boot is not a reqiurement for your test, then I'd suggest to disable it
16:06:36 sean-k-mooney i guess i need to pin artoms code to the older version?
16:06:42 aspiers kashyap: how?
16:06:47 mriedem donnyd: required traits are defined on the flavor or image
16:06:50 artom sean-k-mooney, I hit that a few times when I git pulled a new service version, but the version in the database for the service was old
16:06:56 sean-k-mooney artom: well i didnt use the head of master. i used the version of master your code was bassed on
16:06:58 kashyap aspiers: You mean, "disable how"?
16:07:10 aspiers kashyap: yes, how do I disable it
16:07:13 sean-k-mooney e.g. this is the frist commit before your code
16:07:23 dansmith sean-k-mooney: you'll have to delete some service records
16:07:33 sean-k-mooney ah right
16:07:37 dansmith sean-k-mooney: that message is coming because you've got entries in the db that say everything should be upgraded already
16:08:07 sean-k-mooney if i just stop the agent and do a compute servcie delete that should fix it right
16:08:18 sean-k-mooney alternitivel i can restack that node
16:08:44 donnyd The scenario is I use some sort of discovery to populate placement with traits, and I don't want to configure a flavor for each specific type
16:08:44 donnyd Is there any way to place a custom request in via the sdk?
16:08:53 kashyap aspiers: Don't know top off my head; but let me check with a dev who does this more often
16:09:36 mriedem stephenfin: looking at https://review.opendev.org/#/c/621476/ alex is +2 and you asked for docs fixes in PS61 which are done now, so you might want to hit that again soonish,
16:09:51 kashyap aspiers: In the BIOS menu, it depends on your vendor
16:09:51 mriedem stephenfin: especially b/c i'm +2 on https://review.opendev.org/#/c/673133/15 which is competing for the same microversion
16:10:06 aspiers kashyap: we are talking about a VM in nova here, remember
16:10:16 stephenfin mriedem: Sure, I can do that before I leave
16:10:23 aspiers it's some virtual BIOS thing
16:11:39 kashyap aspiers: Right, it's a VM; a quick check -- you're using one of those enrolled "VARS" files shipped by your distro I guess?
16:12:34 kashyap aspiers: If so, the quickest way to disable it is to just start your guest with a non-enrolled (i.e. not signed by MS) "VARS" file.
16:12:40 aspiers kashyap: right, as per my comment on https://review.opendev.org/#/c/348394/
16:13:00 sean-k-mooney we need to update osc to better support service deleteion
16:13:02 aspiers kashyap: in that final comment there is a list of all the "code" files
16:13:10 aspiers kashyap: which have corresponding "vars" files
16:13:15 aspiers kashyap: which one looks most promising?
16:13:23 kashyap Looking
16:13:33 sean-k-mooney osc only support the id not the service uuid
16:13:53 mriedem sean-k-mooney: that's not accurate
16:13:53 aspiers kashyap: currently I'm using ovmf-x86_64-suse-code.bin and ovmf-x86_64-suse-vars.bin
16:14:06 aspiers kashyap: I have no idea what "4m" means
16:14:06 mriedem https://docs.openstack.org/python-openstackclient/latest/cli/command-objects/compute-service.html#compute-service-delete
16:14:22 kashyap aspiers: I once did research on that SUSE naming, but totally forgot
16:14:37 kashyap aspiers: Hmm, so you _are_ using a non-MS signed one
16:14:38 sean-k-mooney well the help text in the version i have installed says it thakes the id and the uuid did not work
16:14:51 mriedem sean-k-mooney: you have to specify the microversion with osc
16:14:53 mriedem it defaults to 2.1
16:15:07 mriedem so: openstack --os-compute-api-version 2.53 compute service delete <service uuid>
16:15:08 sean-k-mooney oh ya that is a thing
16:15:28 kashyap aspiers: Hint: that "4m" will be described in your RPM spec file; but that's not useful for your immediate prob
16:15:32 sean-k-mooney the help text does not methion that
16:15:42 kashyap aspiers: Try this one: ovmf-x86_64.bin (and the corresponding VARS file)
16:15:53 mriedem sean-k-mooney: b/c you have old osc
16:16:01 mriedem i added that recently
16:16:24 sean-k-mooney ah ok im using 3.19.0
16:16:44 sean-k-mooney its not a big deal anyway
16:17:03 aspiers kashyap: OK but as per my comment on https://review.opendev.org/#/c/348394/ ovmf-x86_64-code.bin did not work, so not confident that ovmf-x86_64.bin will work either
16:17:20 kashyap aspiers: Reading that; I was relying on the file name suggested by Dirk correct :-(
16:17:47 kashyap aspiers: Also, I find it crazy that the SUSE RPM has _so_ many binaries; you need a simple combination:
16:18:28 kashyap aspiers: (a) a non-SB binary and a non-enrolled VARS file; (b) an SB-binary and an enrolled (MS-signed) VARS file
16:18:39 aspiers yeah I have no clue why it's like that
16:19:23 kashyap aspiers: Can you get a link to the RPM spec file, please? That will clarify what they are all.
16:20:37 kashyap aspiers: Incidentally, the "light at the end of the tunnel" is it's going to be much easier, once SUSE ships: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=932269
16:20:39 openstack Debian bug 932269 in ovmf "Ship the firmware "descriptor files" as part of the 'ovmf' package" [Normal,Fixed]
16:21:04 kashyap (Debian and Ubuntu now do. As does Fedora, for a while)
16:21:58 sean-k-mooney hum so i still get the same error after deleting the compute service
16:23:35 aspiers kashyap: https://build.opensuse.org/package/view_file/Virtualization/ovmf/ovmf.spec?expand=1
16:23:44 dansmith sean-k-mooney: you have to stop them all, delete them all, then start up the old one first
16:24:17 sean-k-mooney oh i have stop and delete the compute service on the "new" node too
16:24:25 sean-k-mooney not just the one im puting ot master
16:24:27 sean-k-mooney ok
16:24:40 donnyd mriedem: is there a way to craft custom requests via the sdk?
16:25:03 sean-k-mooney donnyd: if you mean raw http requests yes
16:25:07 aspiers kashyap: do you really think the exception message should refer to "Q35" when the way to fix the exception is to set hw_machine_type=q35 (lowercase) ?
16:26:04 sean-k-mooney donnyd: you can use the sdk to jsut get a rest client and then perform raw GET|POST|PUT|DELETE|... request with the client with whatever body you want
16:26:31 donnyd So is it possible to craft a raw request that would allow me to ask placement for a node that meets X trait and then tell nova to provision that resource
16:26:35 kashyap aspiers: Thanks for the link; that "4m" corresponds to a 4MB flash image.
16:26:42 aspiers ah interesting

Earlier   Later