| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2024-11-07 | |||
| 08:22:29 | opendevreview | Balazs Gibizer proposed openstack/nova stable/2024.2: Route shared storage RPC to evac dest at startup https://review.opendev.org/c/openstack/nova/+/934302 | |
| 08:24:38 | opendevreview | Balazs Gibizer proposed openstack/nova stable/2024.1: Reproduce bug 2085975 in functional https://review.opendev.org/c/openstack/nova/+/934303 | |
| 08:24:39 | opendevreview | Balazs Gibizer proposed openstack/nova stable/2024.1: Route shared storage RPC to evac dest at startup https://review.opendev.org/c/openstack/nova/+/934304 | |
| 08:26:06 | opendevreview | Balazs Gibizer proposed openstack/nova stable/2023.2: Reproduce bug 2085975 in functional https://review.opendev.org/c/openstack/nova/+/934305 | |
| 08:26:07 | opendevreview | Balazs Gibizer proposed openstack/nova stable/2023.2: Route shared storage RPC to evac dest at startup https://review.opendev.org/c/openstack/nova/+/934306 | |
| 08:28:29 | opendevreview | Balazs Gibizer proposed openstack/nova stable/2023.1: Reproduce bug 2085975 in functional https://review.opendev.org/c/openstack/nova/+/934308 | |
| 08:28:30 | opendevreview | Balazs Gibizer proposed openstack/nova stable/2023.1: Route shared storage RPC to evac dest at startup https://review.opendev.org/c/openstack/nova/+/934309 | |
| 09:57:14 | chris218 | Hi guys, how does stuff defined in api/validation/extra_specs translate to virt/libvirt/config? I'm trying to figure out how os_boot_dev is set in LibvirtConfigGuest since it can't be used with boot_order in LibvirtConfigGuestDisk | |
| 10:26:20 | opendevreview | Stephen Finucane proposed openstack/os-vif master: mypy: Add type annotations https://review.opendev.org/c/openstack/os-vif/+/745162 | |
| 10:30:59 | sean-k-mooney | chris218: it mostly does not. api/validation/extra_specs is just chekcing of the extra spec values are valid at the api level to determin if we shoudl reject the request early when you are defining a flavor and setting the extra specs | |
| 10:31:19 | sean-k-mooney | the xml generation is entirly seperate | |
| 10:33:11 | sean-k-mooney | os_boot_dev is not an extra spec or image property, we do not support configing which device is bootable. its either the root disk or the rescue disk | |
| 10:34:12 | chris218 | Yea but stuff like hw:boot_menu then translates to os_bootmenu, right? | |
| 10:34:47 | sean-k-mooney | it can yes, we intentionally do not expose everyting a hypervior can configure | |
| 10:36:59 | sean-k-mooney | chris218: is there a partical reason your interested in modifying the boot dev | |
| 10:37:48 | chris218 | I'm still playing with the idea of exposing network booting through some flavor extra spec | |
| 10:38:19 | sean-k-mooney | ah the last time this came up i suggested adding it as a property on the neutron port | |
| 10:38:33 | chris218 | I see that libvirt supports <boot order="X"> on interfaces | |
| 10:38:34 | sean-k-mooney | basicaly just adding a boot flag to a port | |
| 10:39:00 | chris218 | Yea | |
| 10:39:03 | sean-k-mooney | if you use the boot menu today you can network boot already | |
| 10:39:39 | sean-k-mooney | neuton can set the tftp server info in the dhcp repsonce on the port too | |
| 10:39:40 | chris218 | You'd have to select it manually first though | |
| 10:39:47 | sean-k-mooney | yes you would | |
| 10:40:42 | sean-k-mooney | the thing is nova does not currently suppprot creating a vm wihtout a glance image or a cinder volume. | |
| 10:40:55 | sean-k-mooney | so you would have to deal with that ux as well | |
| 10:41:52 | chris218 | Last time I asked about it somebody suggested using rescue image with it pxe but downside of this is that unrescuing is kind of inflexible and you can't power off and then unrescue server | |
| 10:43:03 | sean-k-mooney | we could supprot unrescue to stopped | |
| 10:43:21 | sean-k-mooney | i belive it curretly returns you to active if the server was active before | |
| 15:33:14 | s3rj1k | Hi all, would this RFA https://bugs.launchpad.net/nova/+bug/2086867 need an actual spec or specless blueprint is enough? | |
| 15:36:54 | gibi | s3rj1k: If it has an API impact, requires DB schema change, or has an upgrade impact then you need a spec. Otherwise it would make sense to propose it as a specless bp and put it on the agenda of some upcoming nova meeting. | |
| 15:37:30 | gibi | and we can discuss it on the meeting if it is simple enough to accept it as specless | |
| 15:40:33 | s3rj1k | gibi: it adds a config option but other than that it's pretty much a small change, can we assume it's specless and in case if it is not I will do spec for it later? After upcoming nova meeting | |
| 15:42:40 | gibi | s3rj1k: yeah, that approach make sense | |
| 15:43:14 | gibi | lets assume it is specless until the meeting discussion | |
| 15:44:49 | s3rj1k | gibi: great thanks, also when/where is the community meeting will take place? | |
| 15:47:16 | s3rj1k | Tuesdays 16:00 UTC (#openstack-nova on OFTC) ? | |
| 15:52:58 | gibi | s3rj1k: correct. https://wiki.openstack.org/wiki/Meetings/Nova | |
| 15:53:26 | gibi | you can add you blueprint to the agenda by editing that page | |
| 15:53:39 | gibi | please add it to the Open discussion section | |
| 15:57:43 | gibi | bauzas: ^^ fyi | |
| 16:02:15 | s3rj1k | gibi: done, thanks for walking me through the formalities | |
| 16:07:46 | sean-k-mooney | im not sure if this need a code change | |
| 16:08:01 | sean-k-mooney | dont we already have the ablity to specify the migration and livemigration inbound adress | |
| 16:08:17 | sean-k-mooney | if you set those to an ip then nova will use the ip for scp too i belive | |
| 16:08:47 | sean-k-mooney | https://docs.openstack.org/nova/latest/configuration/config.html#libvirt.live_migration_inbound_addr and https://docs.openstack.org/nova/latest/configuration/config.html#libvirt.migration_inbound_addr | |
| 16:10:00 | sean-k-mooney | s3rj1k: ^ | |
| 16:40:26 | s3rj1k | sean-k-mooney: I'll try to verify this before meetup, thanks for heads-up | |
| 17:41:43 | sean-k-mooney | bauzas: can you add https://review.opendev.org/q/topic:%22bug/2074219%22 to your review queue in the morning | |
| #openstack-nova - 2024-11-08 | |||
| 01:46:34 | opendevreview | melanie witt proposed openstack/nova-specs master: Re-propose config option for behavior of unset unified limits https://review.opendev.org/c/openstack/nova-specs/+/934391 | |
| 01:54:01 | melwitt | bauzas: just fyi it appears the openstack-tox-docs job on the nova-specs repo is broken. not sure what changed, I'm guessing something with sphinx? | |
| 01:54:36 | sean-k-mooney | melwitt: what python version did you use | |
| 01:55:13 | melwitt | sean-k-mooney: I have 3.10.12 but it's failing in the gate also for the same failure | |
| 01:55:49 | sean-k-mooney | ok im fixing a py3.9 comaptiblity issue with the openstackdoctheme | |
| 01:56:59 | melwitt | sean-k-mooney: oh, nice. it looks like in the gate it is 3.9 https://zuul.opendev.org/t/openstack/build/71de468bcd5b419994f13b8bcff71f16/log/job-output.txt#440 so likely to be fixed with your fix I expect | |
| 01:57:16 | sean-k-mooney | this is not the bug im fixing | |
| 01:57:48 | sean-k-mooney | we build with warnings threaded as errors | |
| 01:57:48 | melwitt | oh | |
| 01:57:50 | sean-k-mooney | https://paste.opendev.org/show/bsKc2Q1VihOwLG5Czvd4/ | |
| 01:58:23 | melwitt | yeah. this just seems recent but I guess I hadn't noticed | |
| 01:58:26 | sean-k-mooney | so ya a new sphinx release happend on Oct 13, 2024 | |
| 01:58:40 | sean-k-mooney | so i think those are just new warnings that should be simple enough to fix | |
| 01:58:57 | melwitt | someone might have already proposed a fix. checking.. | |
| 01:59:56 | melwitt | hm no, doesn't look like it https://review.opendev.org/q/project:openstack/nova-specs+is:open | |
| 01:59:56 | sean-k-mooney | this is my docsthme patch in case your interested https://review.opendev.org/c/openstack/openstackdocstheme/+/934384 | |
| 02:00:44 | melwitt | ah cool | |
| 02:05:37 | sean-k-mooney | im going to go to sleep quickly but ill just confirm i can repodcue the issues locally. do you plan ot fix them or will i? | |
| 02:06:31 | melwitt | sean-k-mooney: up to you. I don't know what the footnote warning means off the top of my head | |
| 02:06:55 | melwitt | so it might be faster if you do it | |
| 02:07:32 | melwitt | hah. that's cause nova-specs renders everything from the beginning of time | |
| 02:08:13 | sean-k-mooney | an with that done its a perfect time to rememeber tha ton py39 we clamp to an older version of sphinx | |
| 02:08:41 | sean-k-mooney | on the plus side is this deffinly just a problem because a new version was released | |
| 02:08:51 | melwitt | goes back to at least juno | |
| 02:10:03 | sean-k-mooney | i vagly remember the before times of the nova wiki | |
| 02:10:24 | sean-k-mooney | it was confusing when they were first being intoduced but it was a big imporvement in the end | |
| 02:10:28 | melwitt | oh yeah | |
| 02:22:18 | sean-k-mooney | so the problem seems to be that ehre are a missmatch in the number of footnotes defiend vs used. | |
| 02:22:42 | sean-k-mooney | we use a footnote by doing [#]_ allows use to not have to name the link | |
| 02:23:04 | sean-k-mooney | we define it by doing .. [#] <linke goes here> | |
| 02:23:25 | sean-k-mooney | if we look at /home/smooney/repos/nova-specs/doc/source/specs/queens/implemented/multi-attach-volume.rst:304: WARNING: Footnote [#] is not referenced. [ref.footnote] | |
| 02:23:39 | sean-k-mooney | there are 4 footnotes but only 3 are used in the doc | |
| 02:24:01 | melwitt | ahhh | |
| 02:24:31 | sean-k-mooney | im going to head to bed but i can take a look in the morning assuming you dont get to it tonight | |
| 02:24:53 | sean-k-mooney | it shoudl be pretty simple to fix we just need to figure out where the extra one shoudl go and or delete it | |
| 02:24:54 | melwitt | ok, seeya tomorrow o/ | |
| 08:25:07 | zigo | I upgraded to latest point release of Nova for both Antelope and Bobcat, in both case, when running unit tests, I get the OOM killer saying hello ... | |
| 08:25:15 | zigo | Any clue what unit test is using so much RAM ? | |
| 08:32:57 | gibi | zigo: interesting we did not noticed that in the upstream gate. What are the memory limits in your env? | |
| 08:33:15 | zigo | I run my jenkins package building with 8 GB of RAM. | |
| 08:33:44 | gibi | thanks I will try to check it in 8G VM then :) | |
| 08:33:46 | zigo | There's Jenkins itself taking a bit of RAM of course, but like 6 GB at least should be available for the unit tests, I'm guessing. | |
| 08:34:07 | zigo | FYI, I've run latest point release exactly, not tip of stable branches. | |
| 08:34:09 | gibi | what level of parallelism do you use in your env? | |
| 08:34:53 | zigo | I have 4 VCPU, so I'm guessing, 4 threads for running tests. | |
| 08:35:43 | zigo | When I look in "top" with the server being idle, I have only 1.2 GB used by my Jenkins env. | |
| 08:35:53 | zigo | So that's 6.8 GB available for building. | |
| 08:37:54 | gibi | ack | |
| 08:47:17 | frickler | zigo: the CI nodes also use only 8GB, so likely some other delta. can you share the complete oom and the test output? I assume this is on bookworm? we could try to test with that in CI instead of noble/jammy | |
| 08:47:43 | zigo | Yes, I'm running on Bookworm, with a chroot & sbuild. | |
| 08:54:15 | zigo | frickler: It killed "java" (ie: my jenkins...), so you wont see anything in the OOM kill log. | |