| opendevreview | melanie witt proposed openstack/nova master: TPM: bump service version to enable live migration https://review.opendev.org/c/openstack/nova/+/975724 | 02:12 |
|---|---|---|
| opendevreview | melanie witt proposed openstack/nova master: TPM: test live migration between hosts with different security https://review.opendev.org/c/openstack/nova/+/952629 | 02:12 |
| opendevreview | melanie witt proposed openstack/nova master: TPM: add late check for supported TPM secret security https://review.opendev.org/c/openstack/nova/+/956975 | 02:12 |
| opendevreview | melanie witt proposed openstack/nova master: TPM: enable conversion of secret security modes via resize https://review.opendev.org/c/openstack/nova/+/962052 | 02:12 |
| opendevreview | melanie witt proposed openstack/nova master: DNM vtpm tempest https://review.opendev.org/c/openstack/nova/+/957477 | 02:12 |
| *** mhen_ is now known as mhen | 02:23 | |
| *** erlon9 is now known as erlon | 03:27 | |
| opendevreview | Merged openstack/nova master: Set openstacksdk-functional-devstack non voting https://review.opendev.org/c/openstack/nova/+/976095 | 07:27 |
| gokhan | sean-k-mooney, Thanks for your previous guidance. It turns out I misunderstood the live_migration_completion_timeout parameter; I thought it represented the total duration for the entire migration process, but I’ve now realized it is calculated per GiB of RAM. After significantly lowering this value, I observed that the post-copy transition is finally being triggered as expected. | 07:56 |
| gokhan | I also wanted to get your advice on network architecture | 07:56 |
| gokhan | Currently, we are using the management network for live migration. Do you recommend separating the live migration traffic onto a dedicated network? In what specific scenarios or traffic loads does isolating this network become a necessity? | 07:56 |
| gokhan | Would increasing the MTU from 1500 to 9000 (Jumbo Frames) provide a significant benefit for live migration performance, especially for VMs with 64GB+ RAM? | 07:57 |
| Callum027 | In our biggest cloud region (~35 hypervisors) the biggest cause of slow live migrations is how busy the instance is, not the capacity of the network. Our instances are migrated across the management network (though we use Ceph RBD for block storage, and that has its own dedicated network with jumbo frames enabled). | 09:05 |
| lajoskatona | sean-k-mooney, Uggla: Hi, there's a bug for IOThreads: https://bugs.launchpad.net/nova/+bug/2140537 , and I already pushed reproducer and a wip fix for it: https://review.opendev.org/q/topic:%22bug/2140537%22 , please comment if my understanding is wrong for how it should work, thanks in advance | 09:59 |
| Uggla | Hi lajoskatona, oh. Thanks. Gonna review it asap. Super you jump on that bug. 👍️ | 10:01 |
| opendevreview | Tobias Urdin proposed openstack/nova master: Add a timeout for `ceph df` command https://review.opendev.org/c/openstack/nova/+/975927 | 10:19 |
| aloke-dev | \help | 10:19 |
| gokhan | thanks for the insight , Callum027 ! I’ve been observing exactly what you described: even with available bandwidth, the high dirty rate of our 64GB+ instances was keeping the migration in a loop. I’ve adjusted our live_migration_completion_timeout (realizing now it's per GiB) to trigger post-copy earlier, which seems to be the only reliable way to 'break' that loop for these busy instances. | 10:36 |
| gibi | lajoskatona: I'm +2 on the reproducer and has some comments in the fix. But overall nice work. Thanks | 10:47 |
| sean-k-mooney | lajoskatona: ack ill take a look shortly | 10:57 |
| sean-k-mooney | i agree the repoducer looks good over all ill take a look at the fix a littel later. | 11:05 |
| lajoskatona | thanks guys, gibi: I check your comments | 11:13 |
| opendevreview | Merged openstack/nova master: Add regression test to repoduce bug 2140537 https://review.opendev.org/c/openstack/nova/+/975832 | 11:35 |
| opendevreview | Stephen Finucane proposed openstack/os-vif master: typing: Add initial type annotations https://review.opendev.org/c/openstack/os-vif/+/745162 | 11:36 |
| opendevreview | Stephen Finucane proposed openstack/os-vif master: pre-commit: Bump versions https://review.opendev.org/c/openstack/os-vif/+/976358 | 11:36 |
| opendevreview | Stephen Finucane proposed openstack/os-vif master: Migrate setup configuration to pyproject.toml https://review.opendev.org/c/openstack/os-vif/+/976359 | 11:36 |
| opendevreview | Stephen Finucane proposed openstack/os-vif master: tests: Remove errant parameters https://review.opendev.org/c/openstack/os-vif/+/976360 | 11:36 |
| opendevreview | Stephen Finucane proposed openstack/os-vif master: trivial: Use bit shifting https://review.opendev.org/c/openstack/os-vif/+/976361 | 11:36 |
| opendevreview | Stephen Finucane proposed openstack/os-vif master: typing: Add typing to vif_plug_noop module https://review.opendev.org/c/openstack/os-vif/+/976362 | 11:36 |
| opendevreview | Stephen Finucane proposed openstack/os-vif master: WIP: typing: Add typing to vif_plug_ovs module https://review.opendev.org/c/openstack/os-vif/+/976363 | 11:36 |
| opendevreview | Stephen Finucane proposed openstack/os-vif master: WIP: typing: Use generics https://review.opendev.org/c/openstack/os-vif/+/976364 | 11:37 |
| opendevreview | Elod Illes proposed openstack/nova stable/2024.2: [CI][stable-only] Fix py312 based jobs https://review.opendev.org/c/openstack/nova/+/976373 | 12:09 |
| obre | sean-k-mooney: I have tested that the os-trait update and an addition to /usr/lib/python3/dist-packages/nova/virt/libvirt/utils.py is enough to let the compute-nodes announce its amx-capabilities as traits; and it is possible to add properties to flavors to actually boot VM's with AMX. So now Im going to commit the patch I have created, and are then wondering what I should tag | 13:17 |
| obre | the commit with. There is a bp for this (https://blueprints.launchpad.net/nova/+spec/add-amx-traits), but is it correct to reference that if this is considered to be a wishlist-bug? I guess the alternative is that I create a bug describing the issue; and then commit a fix for it? Or should I neither add a Closes-Bug nor a Implements in the commit-message? | 13:17 |
| sean-k-mooney | you can use implements its fine | 13:25 |
| sean-k-mooney | we will jsut need to approve the spec in lacuhpad after | 13:25 |
| obre | Ack! | 13:26 |
| *** iurygregory_ is now known as iurygregory | 13:26 | |
| *** haleyb_ is now known as haleyb | 13:36 | |
| opendevreview | Eigil Obrestad proposed openstack/nova master: Make nova recognize amx-capabilities https://review.opendev.org/c/openstack/nova/+/976390 | 13:42 |
| nicolairuckel | Any updates on the CI? | 13:43 |
| obre | sean-k-mooney: Whats missing now is to update the version-requirement for os-traits. Where can I see when a new release is there? And should i create a release-note for this? In that case; whats the naming-convention for files in releasenotes/notes? | 13:44 |
| sean-k-mooney | we need to tag a release via the releases repo then bump the upper-constraint in the requirements repo | 13:44 |
| obre | Do you by "requirements repo" mean the requirements.txt in the nova repo? | 13:47 |
| sean-k-mooney | no i mean we need to raise the max version here https://github.com/openstack/requirements/blob/master/upper-constraints.txt#L347 | 13:50 |
| sean-k-mooney | we have a bot to do that aftger the release is made | 13:50 |
| sean-k-mooney | obre: ill propse the release for you now | 13:57 |
| obre | Looking at the requirements-repo it looks like its bots that updates that as well? It might be me overthinking this; but im currently a bit confused on where to indicate that my nova-patch needs the new os-traits release? | 13:57 |
| obre | sean-k-mooney: Thanks. | 13:57 |
| sean-k-mooney | the new version will be 3.6.0 | 13:57 |
| obre | Is adding a "Depends-On: https://review.opendev.org/c/openstack/os-traits/+/868149" whats needed in 'my' commit-message here so that the bots know what version is needed? Or should I add the version-number somewhere? | 14:01 |
| sean-k-mooney | you will need to update requirements.txt to raise the min version | 14:02 |
| sean-k-mooney | that will initally cause the jobs to fail until the release happens and the max is raised in the requirements repo | 14:02 |
| sean-k-mooney | but we can just recheck it when all the deps are in place | 14:03 |
| obre | Ok. Then Ill update the requirements.txt and check back later to see if the release is through and then ask someone to trigger a re-check? | 14:03 |
| sean-k-mooney | https://review.opendev.org/c/openstack/releases/+/976392 is the release patch Uggla bauzas can ye appoved that when you have time | 14:04 |
| opendevreview | Eigil Obrestad proposed openstack/nova master: Make nova recognize amx-capabilities https://review.opendev.org/c/openstack/nova/+/976390 | 14:05 |
| Uggla | sean-k-mooney, I'll have a look at it asap. | 14:05 |
| opendevreview | Stephen Finucane proposed openstack/os-vif master: Migrate setup configuration to pyproject.toml https://review.opendev.org/c/openstack/os-vif/+/976359 | 15:16 |
| opendevreview | Stephen Finucane proposed openstack/os-vif master: tests: Remove errant parameters https://review.opendev.org/c/openstack/os-vif/+/976360 | 15:16 |
| opendevreview | Stephen Finucane proposed openstack/os-vif master: trivial: Use bit shifting https://review.opendev.org/c/openstack/os-vif/+/976361 | 15:16 |
| opendevreview | Stephen Finucane proposed openstack/os-vif master: typing: Add initial type annotations https://review.opendev.org/c/openstack/os-vif/+/745162 | 15:16 |
| opendevreview | Stephen Finucane proposed openstack/os-vif master: typing: Add typing to vif_plug_noop module https://review.opendev.org/c/openstack/os-vif/+/976362 | 15:16 |
| opendevreview | Stephen Finucane proposed openstack/os-vif master: typing: Add typing to vif_plug_noop module https://review.opendev.org/c/openstack/os-vif/+/976362 | 15:17 |
| gmaan | nova-cores this is merged means nova gate should be fine now, https://review.opendev.org/c/openstack/nova/+/976095 | 15:34 |
| opendevreview | Ghanshyam proposed openstack/nova master: Revert "Set openstacksdk-functional-devstack non voting" https://review.opendev.org/c/openstack/nova/+/976489 | 15:36 |
| gmaan | sdk fix is also merged, so creating the revert | 15:36 |
| sean-k-mooney | lets see fi it passes but cool we can revert if its green | 16:00 |
| gibi | I have the revert | 16:17 |
| gibi | ssh-keygen -f '/home/gibi/.ssh/known_hosts' -R 'osp-dev-01.lab.eng.brq2.redhat.com' | 16:17 |
| gibi | not that | 16:17 |
| gibi | https://review.opendev.org/c/openstack/nova/+/976096/1 | 16:17 |
| bauzas | did the +2 and the recheck | 16:18 |
| gmaan | gibi: perfect, i abandon mine | 16:21 |
| gibi | ack | 16:30 |
| stephenfin | gibi__: TIL (the ssh-keygen thing). I've always done that manually | 16:57 |
| sean-k-mooney | stephenfin: really? it props you when you need to do it | 17:51 |
| sean-k-mooney | *prompts | 17:51 |
| opendevreview | Merged openstack/nova master: Revert "Set openstacksdk-functional-devstack non voting" https://review.opendev.org/c/openstack/nova/+/976096 | 20:09 |
| melwitt | hm, looks like we might have a bit of a racy unit test in nova-tox-py312-threading: test_submit_second_while_delaying_first AssertionError: 1.9995737350000127 not greater than 2.0 9f820e554411581c95ab2e5aaab45/testr_results.html | 20:48 |
| melwitt | * https://f82174203394791b8a26-ea1d911c7c7db668a9aa6765a743313b.ssl.cf2.rackcdn.com/openstack/7e59f820e55 | 20:48 |
| melwitt | https://f82174203394791b8a26-ea1d911c7c7db668a9aa6765a743313b.ssl.cf2.rackcdn.com/openstack/7e59f820e554411581c95ab2e5aaab45/testr_results.html | 20:52 |
| * melwitt sighs | 20:52 | |
| sean-k-mooney | ya. we normally mock our tiem in the unit tests to some degree | 20:52 |
| sean-k-mooney | there is an assert almost equal we can use for float to have a littel wiggle room | 20:53 |
| opendevreview | Brian Rosmaita proposed openstack/nova master: DNM: run nova-ceph-multistore on debian-trixie https://review.opendev.org/c/openstack/nova/+/976551 | 21:22 |
Generated by irclog2html.py 4.0.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!