| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2024-10-23 | |||
| 15:27:30 | opendevreview | Takashi Kajinami proposed openstack/nova master: libvirt: Launch instances with SEV-ES memory encryption https://review.opendev.org/c/openstack/nova/+/926106 | |
| 15:28:28 | lajoskatona | bauzas: Hi, I am on the Neutron room, so in worst case, but I liked the Friday time if that still can work | |
| 15:28:45 | bauzas | noted, we'll discuss this on Friday then | |
| 15:29:12 | lajoskatona | bauzas: thanks | |
| 15:49:04 | bauzas | tmazur: as a reminder, horizon-nova x-p session at 1600UTC in the austin (nova) room | |
| 15:50:12 | tmazur | bauzas, thanks! We will come. Can be late for a couple minutes since we have another cross project session now | |
| 15:51:28 | bauzas | noted | |
| 21:28:20 | atmark | hello, which table in the db is libvirt xml definition stored? | |
| 22:30:37 | melwitt | atmark: libvirt xml definition are not stored in the db. they are not stored anywhere, they are generated on the fly | |
| 22:58:50 | atmark | melwitt: does that mean I can edit xml file without nova overwritting it? I have issue an with live migrating an instance, libvirt thinks the disk.config of instance is stored locally where in fact it's in ceph https://paste.openstack.org/show/b1ABMBlvqzKQAZtvYnGc/ | |
| 23:00:23 | melwitt | atmark: no, kind of the opposite. nova can overwrite it at certain times. you can temporarily edit it and it will stay until something triggers an xml regeneration, like a hard reboot of the guest for example | |
| 23:02:22 | melwitt | if you're trying to unwedge something it could work to edit the xml and do the thing you're trying to do but the root cause of why it's getting the wrong disk.config location would have to be solved for the issue to be really fixed | |
| 23:04:18 | atmark | nova has to get the libvirt definitions somewhere to generate xml, right ? | |
| 23:04:54 | melwitt | IIRC disk.config format/location is determined by the CONF.libvirt.images_type on the host, if it's 'rbd' it should generate xml for rbd. I'm not sure if you have a deployment with a mix of different CONF.libvirt.images_type on different compute hosts | |
| 23:06:00 | melwitt | it generates the definitions using a variety of data. if you're curious all the code is in nova/virt/libvirt/driver.py in get_*_xml methods | |
| 23:08:23 | melwitt | and all of the xml elements are represented as classes in nova/virt/libvirt/config.py | |
| 23:11:32 | atmark | So this host used to have `images_type = raw` which tells to run VM locally and after a while decided mind change to `images_type = rbd`. Do we need to rebuild the VM in order to switch definition to rbd? | |
| 23:12:41 | atmark | after a while decided to change it `images_type = rbd`* | |
| 23:13:01 | atmark | I'll look into the code | |
| 23:13:03 | atmark | Thanks | |
| 23:13:05 | melwitt | ah ok, that explains it. you may be able to fix it by doing a 'server reboot --hard'. that's the go-to for regenerating xml. as long as nova will re-read the CONF to generate, it should work, but I don't remember if it will re-read or not. I think it should but not 100% sure | |
| 23:13:33 | melwitt | *re-read the conf for determining the config drive format | |
| 23:14:32 | atmark | I'll give that try | |
| 23:14:34 | atmark | Thanks | |
| #openstack-nova - 2024-10-24 | |||
| 01:24:30 | opendevreview | Takashi Kajinami proposed openstack/placement master: Skip functional tests on pre-commit-config.yaml update https://review.opendev.org/c/openstack/placement/+/933191 | |
| 01:24:47 | opendevreview | Takashi Kajinami proposed openstack/placement master: Skip functional tests on pre-commit config update https://review.opendev.org/c/openstack/placement/+/933191 | |
| 01:28:21 | opendevreview | Takashi Kajinami proposed openstack/nova master: Skip functional tests on pre-commit config update https://review.opendev.org/c/openstack/nova/+/933192 | |
| 01:41:05 | opendevreview | Takashi Kajinami proposed openstack/placement master: Skip functional tests on pre-commit config update https://review.opendev.org/c/openstack/placement/+/933191 | |
| 01:41:27 | opendevreview | Takashi Kajinami proposed openstack/placement master: Skip functional tests on pre-commit config update https://review.opendev.org/c/openstack/placement/+/933191 | |
| 01:46:58 | opendevreview | Takashi Kajinami proposed openstack/nova master: Skip functional tests on pre-commit config update https://review.opendev.org/c/openstack/nova/+/933192 | |
| 02:51:47 | sean-k-mooney | atmark: we dont actully supprot changign the image_type on an exsiting comptue node and there is no officall supprot form moving a vm form one storage backend to annother | |
| 02:51:59 | sean-k-mooney | it can be doen in some cases via shelve and unshleve. | |
| 03:08:23 | songwenping | sean-k-mooney, hi, the bug 2059809 doesnot effect V and previsous versions? | |
| 03:11:22 | sean-k-mooney | https://bugs.launchpad.net/nova/+bug/2059809 | |
| 03:11:38 | sean-k-mooney | it would if you had a new enough qemu | |
| 03:12:30 | sean-k-mooney | the underlying vulnerbality was in qemu, openstakc in genrela did not have any protection to prevetn unsafe images being passed to qemu-img | |
| 03:12:55 | sean-k-mooney | if you have a qemu that is unpatched and supprots external datafiles | |
| 03:13:25 | sean-k-mooney | adn you use that with a nova/glance/cinder that is unpatched then your cloud is vulnerable | |
| 03:13:57 | sean-k-mooney | V and previous versoin are end of life or unmainteded | |
| 03:14:33 | sean-k-mooney | so the core team does not backport CVE fixes to them regardless of how severe they are | |
| 03:15:56 | songwenping | i find the bug cherry-picked to W, W is also in unmaintained. | |
| 03:16:49 | sean-k-mooney | yep, there is a seprate unmained branch team | |
| 03:18:32 | sean-k-mooney | although im not sure that was cherry picked usptram to wallaby | |
| 03:18:46 | sean-k-mooney | it was done downstream in some distors to older branches | |
| 03:19:30 | songwenping | where i can find the version correspondence of nova and qemu? | |
| 03:20:11 | sean-k-mooney | https://docs.openstack.org/nova/latest/reference/libvirt-distro-support-matrix.html#min-libvirt-qemu-version-and-next-min-libvirt-qemu-version-table | |
| 03:20:35 | sean-k-mooney | V requried qemu 4.0.0 and libvirt 5.0.0 | |
| 03:24:24 | songwenping | our product use qemu 4.2.0 and libvirt 6.6.0 , but cannot reproduce this bug. | |
| 03:25:13 | sean-k-mooney | it was added in qemu 4.0.0 https://github.com/qemu/qemu/commit/65a3d073e9766ed878474525118dbf0c7233caab | |
| 11:14:40 | opendevreview | Balazs Gibizer proposed openstack/nova master: [libvirt]Support hw_vif_model = igb https://review.opendev.org/c/openstack/nova/+/928584 | |
| 11:14:41 | opendevreview | Balazs Gibizer proposed openstack/nova master: [doc]Developer doc about PCI and SRIOV testing https://review.opendev.org/c/openstack/nova/+/928834 | |
| 11:15:18 | opendevreview | Balazs Gibizer proposed openstack/nova master: Refactor obj_make_compatible to reduce complexity https://review.opendev.org/c/openstack/nova/+/928590 | |
| 11:15:19 | opendevreview | Balazs Gibizer proposed openstack/nova master: [ovo]Add igb value to hw_vif_model image property https://review.opendev.org/c/openstack/nova/+/928456 | |
| 11:15:19 | opendevreview | Balazs Gibizer proposed openstack/nova master: [libvirt]Support hw_vif_model = igb https://review.opendev.org/c/openstack/nova/+/928584 | |
| 11:15:20 | opendevreview | Balazs Gibizer proposed openstack/nova master: [doc]Developer doc about PCI and SRIOV testing https://review.opendev.org/c/openstack/nova/+/928834 | |
| 11:18:18 | gibi | tkajinam: I respined the igb series based on you comments ^^ | |
| 11:21:22 | sean-k-mooney | i didnt get to it last week but i can try kiciking it off in the background today | |
| 11:26:46 | gibi | sean-k-mooney: thanks | |
| 12:30:13 | tkajinam | gibi, thanks ! | |
| 13:02:00 | bauzas | nova sessions are starting now | |
| 13:02:51 | bauzas | sean-k-mooney: gibi: joining ? | |
| 13:03:08 | sean-k-mooney | sure but i have to be late or it would not be me :) | |
| 13:36:18 | dansmith | bauzas: it looks like the move of the imagebackend topic might have landed on top of the equivalent glance topic | |
| 13:36:45 | dansmith | in 30m I need to do that in glance | |
| 13:36:52 | dansmith | and I don't want to miss eventlet either of course ;) | |
| 13:37:00 | bauzas | hah | |
| 13:37:09 | bauzas | today was around both topics :) | |
| 13:37:28 | dansmith | also image encryption is after image formats, which melwitt and I should both be in, until 1530 | |
| 13:37:53 | dansmith | glance is very specific about its scheduling, which is nice, since other people are coming to these topics, | |
| 13:37:54 | bauzas | image encryption topic on glance sessions ? | |
| 13:38:07 | dansmith | with nova not strict it's hard to know, but I assume we'll go over both | |
| 13:38:20 | dansmith | yes, for glance-cinder-and-nova encryption stuffm | |
| 13:38:45 | dansmith | and since we asked them to redefine their stuff for us, and the image format stuff will impact, we should be there | |
| 13:39:04 | bauzas | so, to understand correctly, you and mel wouldn't be able to be in the nova sessions before 1530UTC, right ? | |
| 13:39:23 | dansmith | yeah | |
| 13:39:45 | bauzas | given we have cinder-nova x-p session at 1600UTC, that basically means you both are off nova that day | |
| 13:40:05 | dansmith | oh, I didn't see a time for the cinder one | |
| 13:40:07 | dansmith | ugh | |
| 13:40:09 | bauzas | no worries, I understand but then I need to find proper time for discussing the image backedn thinb | |
| 13:42:04 | bauzas | I'll try to see if I can advance the cinder x-p session | |
| 13:42:13 | dansmith | that would be good | |
| 13:43:12 | opendevreview | Mina Shahsavan pour proposed openstack/nova master: nova-api: Prevent os-resetState API action on instances in disallowed VM states. https://review.opendev.org/c/openstack/nova/+/933311 | |
| 13:48:23 | bauzas | dansmith: the cinder-nova x-p session will be moved to 1400-1430-ish | |
| 13:48:31 | dansmith | \o/ | |
| 13:48:51 | dansmith | as long as we have rosmaita in my session at 1400 since we need him :) | |
| 13:48:57 | bauzas | I propose to keep the eventlet discussion to be discussed at 1530 (should be quick) | |
| 13:49:12 | bauzas | this is just a wrap-up confirming what we need to do | |
| 13:49:24 | bauzas | and then imagebackend discussion for ending the day | |
| 14:21:22 | bauzas | zigo: we may discuss your py3.13 topic today at the PTG, interested in joining ? | |
| 14:52:35 | bauzas | eek, got a network issue apparently | |
| 15:00:59 | johnthetubaguy | @bauzas: I had the same, but it recovered just now. | |
| 15:02:37 | bauzas | dditto | |
| 15:08:21 | opendevreview | Takashi Kajinami proposed openstack/nova master: libvirt: Use native DictOpt for disk_cachemode https://review.opendev.org/c/openstack/nova/+/915390 | |
| 15:10:32 | ralonsoh | sean-k-mooney, what time is tomorrow the Nova eventlet discussion? | |
| 15:10:50 | ralonsoh | (because is tomorrow, right?) | |
| 15:11:16 | opendevreview | Takashi Kajinami proposed openstack/nova master: libvirt: Validate valid cache mode at config input layer https://review.opendev.org/c/openstack/nova/+/933327 | |
| 15:32:24 | bauzas | ralonsoh: this is now | |
| 15:32:32 | ralonsoh | bauzas, I'm in, thanks | |
| 15:32:58 | bauzas | dansmith: melwitt: fyi, we're starting the eventlet discussion now | |
| 15:33:09 | dansmith | we will be done soon I think | |