Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2019-09-09
15:30:40 sean-k-mooney yep
15:30:55 sean-k-mooney the only thing that will use that are things with hw:cpu_policy=dedicated
15:30:59 artom Oh, right, that's how we end up with instances with that resource
15:31:02 mriedem hmm, is that just libvirt specific or actually codified into how the code outside of the driver works?
15:31:05 artom By doing anything NUMA-y
15:31:08 sean-k-mooney so all things with PCPUs will have numa
15:31:26 sean-k-mooney mriedem: technically just libvirt but no other dirver currely support pinning
15:31:58 sean-k-mooney mriedem: the api does not require the guest to have a numa toplogy of 1
15:32:05 sean-k-mooney so people should not rely on it
15:32:13 sean-k-mooney but that is what the only implmentaion does
15:34:32 stephenfin sean-k-mooney: bikeshedding time. I need a name for this variable https://review.opendev.org/#/c/674895/30/nova/virt/libvirt/driver.py@7439
15:35:01 stephenfin I can't use allocations because that conflicts with an existing variable, and instance_allocations doesn't work since we're tracking migrations for both instances and migration records
15:35:17 stephenfin instance_migration_allocations is too ugly
15:36:05 efried bad_allocs?
15:36:34 efried although 'bad' isn't really a good adjective for them, in instances_with_bad_allocations either.
15:37:05 stephenfin yeah, I've renamed the latter to 'allocations_needing_reshape'
15:37:07 efried but instances_with_allocations_that_need_to_be_reshaped is unwieldy, as I'm sure you realized while you were working on this.
15:38:11 efried stephenfin: btw, you've noticed there's a tox failure somewhere down the series yah?
15:38:25 stephenfin Damn. I had not
15:39:11 efried looks like https://review.opendev.org/#/c/680983/1
15:39:31 efried stephenfin: at least ^
15:40:12 artom stephenfin, "legacy_instances"?
15:40:32 stephenfin artom: The list also contains UUIDs of migration records
15:40:51 stephenfin so not quite right either
15:41:01 stephenfin think I'll just stick '_' at the end and be done :)
15:41:22 artom What, what are you trying to find a name for? instances_with_bad_allocations?
15:42:19 artom "legacy_consumers", then?
15:42:36 stephenfin oh, legacy_consumers it is
15:42:50 bauzas stephenfin: a bit late but I do propose 'consumer_allocations' for your naming bikeshed
15:45:20 artom mriedem, so, for my own information, did you get what you wanted out of sean-k-mooney, or do you need his "official" vote on that review?
15:46:00 bauzas stephenfin: looks like something is fscked up https://c4d77360c4f137c70770-625a0eb0440aa527fbdb216e8991f5a6.ssl.cf1.rackcdn.com/680108/4/check/openstack-tox-py27/2b51adb/testr_results.html.gz
15:47:09 bauzas stephenfin: I guess you have to modify the conf opt for vcpu_pin_set somewhere obviously :)
15:49:29 stephenfin bauzas: sec
15:49:48 openstackgerrit Stephen Finucane proposed openstack/nova master: Validate CPU config options against running instances https://review.opendev.org/680107
15:49:48 openstackgerrit Stephen Finucane proposed openstack/nova master: Interim follow ups for cpu-resources series https://review.opendev.org/680983
15:49:49 openstackgerrit Stephen Finucane proposed openstack/nova master: objects: Add 'NUMACell.pcpuset' field https://review.opendev.org/680108
15:49:49 openstackgerrit Stephen Finucane proposed openstack/nova master: trivial: Use sane indent https://review.opendev.org/680229
15:49:50 openstackgerrit Stephen Finucane proposed openstack/nova master: hardware: Differentiate between shared and dedicated CPUs https://review.opendev.org/671800
15:49:51 openstackgerrit Stephen Finucane proposed openstack/nova master: Add support for translating CPU policy extra specs, image meta https://review.opendev.org/671801
15:49:51 openstackgerrit Stephen Finucane proposed openstack/nova master: libvirt: Start reporting 'HW_CPU_HYPERTHREADING' trait https://review.opendev.org/675571
15:49:52 openstackgerrit Stephen Finucane proposed openstack/nova master: Add reshaper for PCPU https://review.opendev.org/674895
15:50:14 mriedem artom: i want an ack on the patch
15:50:23 mriedem to show he's actually looked at your code change
15:50:45 bauzas stephenfin: oh heh
15:52:55 artom mriedem, understood
15:54:27 bauzas gibi: still looking at the cpu-resource series but you're next :)
15:54:46 bauzas not sure I'll have time before EOB, but tomorrow you're up :)
15:54:55 gibi bauzas: no worries. I will end my day soon too
15:55:56 bauzas gibi: the check pipeline is a bit late but if tomorrow we can send you some patches, it should be on time
15:56:20 gibi fingers crossed :)
15:58:12 openstackgerrit Merged openstack/nova master: doc: Fix a broken reference link https://review.opendev.org/680892
16:00:38 mriedem maybe melwitt can jump into that series
16:00:49 mriedem those bottom 6 or so changes are min version validation and functional testing
16:00:53 mriedem so there isn't a ton of context required
16:01:13 mriedem losing a day this week with the gate backed up likely means missing feature freeze
16:01:53 bauzas mriedem: I can work a bit overlate
16:02:26 mriedem or we can have more cores reviewing more code for FF
16:02:52 melwitt gibi's series? I can help
16:03:10 mriedem yeah, the bottom 6 changes are +2'ed,
16:03:20 gibi melwitt: series starts here https://review.opendev.org/#/c/680394
16:03:22 mriedem the runtime code is just about min compute service version compat
16:03:26 gibi melwitt: thanks in advance
16:03:28 melwitt ok, thanks
16:03:32 mriedem and then there are functional test changes on top of those
16:04:16 mriedem melwitt: for context, compute rpc api 5.2 (service version 39) is what enables the compute side stuff https://review.opendev.org/#/c/655721/18/nova/objects/service.py
16:04:22 mriedem so that's what those min compute service version checks are looking for
16:05:49 melwitt cool thanks
16:06:31 zbr any chance to get pre-commit merged? https://review.opendev.org/#/c/665518/
16:16:46 sean-k-mooney mriedem: artom i think donnyd said this earlier but FN is set to 0 capastiy currently as donnyd was having electical work done over the weekend. so the numa jobs are currenly not going to run
16:17:52 mriedem sean-k-mooney: couldn't we use vexxhost?
16:18:51 artom mriedem, sean-k-mooney, FWIW, the passing job ran with the allocations_cpu_tuning stuff
16:19:30 artom I did the recheck at Sep 7 9:52 AM
16:19:50 sean-k-mooney mriedem im going to reach out to mnaser to see if we can in the future but we need multi numa flaovr whic i dont think vexxhost currenly provide to the gate
16:20:08 artom Having uploaded PS60 of https://review.opendev.org/#/c/635229/60/ at Sep 7 9:26 AM
16:20:16 mriedem artom: did you confirm things in the logs as you'd expect?
16:20:28 mriedem like xml you'd expect to see?
16:20:30 sean-k-mooney mriedem: if vexxhost would be will to add one like the ones FN have added then yes we could use both vexxhost and FN for this
16:20:32 artom The changes since then are basically comments: https://review.opendev.org/#/c/635229/60..61/nova/tests/unit/virt/libvirt/test_driver.py
16:20:48 artom mriedem, yep, I even added debug logging to the _update_numa_xml method
16:20:54 mriedem artom: yeah i saw and commented on PS61
16:22:02 sean-k-mooney mnaser: not to put you on the spot but is creating a custom flavor for the upstream ci to consume possible/something you/vexxhost would be willing to consider?
16:22:15 artom So for example input: https://zuul.opendev.org/t/openstack/build/dd0f5dad770d40a2afb3c506327d1b3e/log/compute/logs/screen-n-cpu.txt.gz#2081
16:22:24 sean-k-mooney mnaser: i can provide the detail later if so.
16:22:34 artom And then output: https://zuul.opendev.org/t/openstack/build/dd0f5dad770d40a2afb3c506327d1b3e/log/compute/logs/screen-n-cpu.txt.gz#2195
16:23:17 artom We see the vcpupin go from 0-3 and 4-6 to 1-3 and 4-7
16:23:49 artom And then for the allocations_cpu_tuning, I have https://review.opendev.org/#/c/680806/
16:24:00 openstackgerrit Chris Dent proposed openstack/nova master: Don't duplicate PlacementFixture in libvirt func tests https://review.opendev.org/681059
16:24:04 cdent mriedem, efried ^
16:24:20 artom Again, input: https://zuul.opendev.org/t/openstack/build/f79144ae28114cea96e708d0bb5ebfb1/log/logs/screen-n-cpu.txt.gz#2565
16:24:30 artom And output: https://zuul.opendev.org/t/openstack/build/f79144ae28114cea96e708d0bb5ebfb1/log/logs/screen-n-cpu.txt.gz#2666
16:24:33 donnyd sean-k-mooney: artom correct FN is at zero for the moment
16:24:50 artom <vcpu placement="static" cpuset="0">1</vcpu> in both cases
16:25:02 artom Because I couldn't figure out how to set vcpu_pin_set per-node
16:25:08 artom But at least it runs through the code
16:25:09 donnyd The electrical work is all done.... Now just waiting on propane and FN will be back in business
16:25:28 sean-k-mooney artom: i have set vcpu pin set per node in one of them
16:25:59 sean-k-mooney artom: https://review.opendev.org/#/c/679805/
16:26:18 sean-k-mooney artom: and that also passed on saturday
16:26:27 artom sean-k-mooney, what's with a NUMA flavor though, right?
16:26:33 mriedem artom: ok can you link those into the patch for reference later?

Earlier   Later