| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2022-04-25 | |||
| 14:38:27 | gibi | the reason was that I started with teh cleanup path as it was easier to get to it. but no technical dependency so I can reorder if needed | |
| 14:39:51 | sean-k-mooney | no thats ok in principal once the start up check is merged you dont need the others | |
| 14:40:00 | sean-k-mooney | so that is why i was suggesting reordering | |
| 14:40:04 | sean-k-mooney | but i like having all 3 | |
| 14:40:27 | sean-k-mooney | (you would have to restart the agent to get this code hench the startup cleanup is enough) | |
| 14:42:02 | gibi | the current cleanup logic only works if the admin removes the VFs from the hypervisor first, then restart the agent. So if that removal (of the restart) is not possible then the claim-while-child-unavailable patch helps | |
| 14:42:14 | sean-k-mooney | ah you are right | |
| 14:42:35 | gibi | or I could create a wider cleanup logic that does not require the VF removal from the hypervisor | |
| 14:42:55 | gibi | just push the VF to available at startup | |
| 14:43:01 | gibi | regardles if it is removed or not | |
| 14:43:25 | gibi | and do this before the existing removed device cleanup logic | |
| 14:43:39 | gibi | we have full of options :) | |
| 14:45:35 | sean-k-mooney | im reviewing it now so unless others object lets keep this simple and go with what you have | |
| 14:45:48 | sean-k-mooney | we might want to do a larger cleanup later but this looks pretty backportable to me | |
| 14:46:33 | gibi | yepp it is pretty self contains now | |
| 14:46:50 | gibi | I will do the backports too | |
| 14:46:56 | gibi | once it lands on master | |
| 14:50:27 | opendevreview | Artom Lifshitz proposed openstack/nova master: Reproduce live migration rollback w/o multi port bindings error https://review.opendev.org/c/openstack/nova/+/839091 | |
| 14:50:28 | opendevreview | Artom Lifshitz proposed openstack/nova master: Fix LM rollback w/o multi port bindings extension https://review.opendev.org/c/openstack/nova/+/839227 | |
| 15:56:11 | opendevreview | Andre Aranha proposed openstack/nova master: Test setting the nova job to centos-9-stream https://review.opendev.org/c/openstack/nova/+/831844 | |
| 16:13:57 | sean-k-mooney | ah fips | |
| 16:14:02 | sean-k-mooney | i was like which job | |
| 16:30:41 | opendevreview | Dan Smith proposed openstack/nova master: DNM: Run against performance.json patch https://review.opendev.org/c/openstack/nova/+/838934 | |
| 17:45:11 | melwitt | gmann: this is a small patch involving deprecation of [api] config option that you might be interested in reviewing https://review.opendev.org/c/openstack/nova/+/836253 | |
| 18:05:45 | gmann | melwitt: thanks, +A. lgtm. | |
| 18:13:54 | melwitt | thanks gmann | |
| 18:32:40 | opendevreview | Jakub Libosvar proposed openstack/os-vif master: Check for hybrid plugging in OVS https://review.opendev.org/c/openstack/os-vif/+/839102 | |
| 19:39:55 | opendevreview | Merged openstack/nova master: Deprecate [api] use_forwarded_for https://review.opendev.org/c/openstack/nova/+/836253 | |
| 23:15:07 | opendevreview | Ghanshyam proposed openstack/nova master: Move centos stream testing to centos-9-stream https://review.opendev.org/c/openstack/nova/+/839275 | |
| #openstack-nova - 2022-04-26 | |||
| 01:00:32 | fzzf[m] | hi folks. I use nodepool connect openstack provider. in openstack, nova build instance failed. and nova list show instance is in BUILD state. and I can't delete it.... (full message at https://matrix.org/_matrix/media/r0/download/matrix.org/JmBToOhokPkttSFIalLvbSEy) | |
| 01:33:44 | opendevreview | Jorhson Deng proposed openstack/nova master: Reset the instance's task_state to None when the status is error https://review.opendev.org/c/openstack/nova/+/838991 | |
| 03:28:53 | fzzf[m] | * hi folks. I use nodepool connect openstack provider. in openstack, nova build instance failed. and nova list show instance is in BUILD state. and I can't delete it.... (full message at https://matrix.org/_matrix/media/r0/download/matrix.org/nhRIVjZthAjNvxgkNkdUczzV) | |
| 07:18:08 | Uggla | Good morning. o/ | |
| 07:37:56 | bauzas | Uggla: good morning | |
| 08:31:04 | gibi | o/ | |
| 08:56:04 | sean-k-mooney | gmann: dansmith care to readd your +w for moving the fips job to c9s https://review.opendev.org/c/openstack/nova/+/831844/7 | |
| 09:30:24 | fzzf[m] | hi folks. I use nodepool connect openstack provider. nova build instance failed. and nova list show instance is in BUILD state like follow. I delete it prompt No server with a name or ID. and in dashboard show have use these two instance, 8 vcpu, 16g memory, openstack is packstack train ,nova is 15.1.1... (full message at https://matrix.org/_matrix/media/r0/download/matrix.org/cXVvKGZbhOphiZZcinYshnCe) | |
| 10:30:31 | opendevreview | Balazs Gibizer proposed openstack/nova stable/train: Reproduce bug 1953359 https://review.opendev.org/c/openstack/nova/+/839353 | |
| 10:30:32 | opendevreview | Balazs Gibizer proposed openstack/nova stable/train: Extend the reproducer for 1953359 and 1952915 https://review.opendev.org/c/openstack/nova/+/839354 | |
| 10:30:33 | opendevreview | Balazs Gibizer proposed openstack/nova stable/train: [rt] Apply migration context for incoming migrations https://review.opendev.org/c/openstack/nova/+/839355 | |
| 11:03:07 | gibi | bauzas: I'm not when you clean up the nova meeting agenda but I now added a topic for Open Discussion for this week's meeting | |
| 11:03:14 | gibi | * I'm not sure | |
| 11:10:46 | sean-k-mooney | artom: by the way while you are fixing revert if you feel like updating https://review.opendev.org/c/openstack/nova/+/750217/2 i would not object. i should really get back to that but have not had the time | |
| 11:22:41 | opendevreview | Andrew Bonney proposed openstack/nova master: Fix segment-aware scheduling permissions error https://review.opendev.org/c/openstack/nova/+/839361 | |
| 11:57:36 | sean-k-mooney | woot! just tested cold migration with vdpa on real hardware with my WIP patch and it indeed worked as the func tests said it should | |
| 11:58:15 | sean-k-mooney | i need to test every thing else and then reivew the logs but this is a good sign | |
| 12:01:23 | sean-k-mooney | shelve and resize work too | |
| 12:09:39 | sean-k-mooney | evacuate works too and the device is correctly released if/when the source host is started after the evac | |
| 12:10:11 | sean-k-mooney | so that is all the move ops that should work exluding live-migration which qemu does not support | |
| 12:12:36 | sean-k-mooney | ok suspend is broken but that fine that is going to be split into another patch anyway and i am pretty sure i know why that is broken anyway. im using the generic pci device detach function and i shoudl be using the interface detach version since vdpa devices ues the interface not host dev elements like normal sriov vfs | |
| 12:14:14 | gibi | sean-k-mooney: nice! | |
| 12:16:14 | sean-k-mooney | im really happy i was able to more or less prove this out with the functional tests ahead of time | |
| 12:16:34 | sean-k-mooney | i kindof expected suspend to fail too which is why i wanted to test this on real hardware | |
| 12:25:11 | Uggla | sean-k-mooney, I guess you manage to install ceph on top of Centos Stream. Any hints to fix that: nothing provides "libpython3.6m.so.1.0()(64bit) needed by ceph-mgr-2:16.2.7-0.el8.x86_64" | |
| 12:32:30 | sean-k-mooney | Uggla: how are you isntalling ceph? with the devstack plugin or by hand? | |
| 12:33:31 | sean-k-mooney | by the way that looks like you are just missin python3-devel | |
| 12:34:02 | sean-k-mooney | also are you using centos stream 8 | |
| 12:34:07 | sean-k-mooney | that is not supported anymore | |
| 12:34:15 | Uggla | sean-k-mooney, devstack plugin (from manila local.conf example : https://docs.openstack.org/manila/latest/_downloads/e31f39ce83daac60c4348fde522cee1e/cephfs_local.conf) | |
| 12:34:18 | sean-k-mooney | you should be using centos stream 9 | |
| 12:34:35 | sean-k-mooney | openstack nolonger support python 3.6 | |
| 12:34:35 | Uggla | sean-k-mooney, it is a stream9 | |
| 12:34:54 | sean-k-mooney | libpython3.6 is python 3.6 | |
| 12:35:01 | sean-k-mooney | centos stream 9 uses 3.9 | |
| 12:35:21 | Uggla | for anyreason it seems the plugin is trying to install a pkg from centos 8 | |
| 12:37:25 | Uggla | Problem: package ceph-2:16.2.7-0.el8.x86_64 requires ceph-mgr = 2:16.2.7-0.el8 but it is clearly a stream 9. I'm gonna have a look where the ceph package comes from. | |
| 12:38:18 | sean-k-mooney | i dont think they have added centos stream 9 support to the devstack plugin yet | |
| 12:38:25 | sean-k-mooney | just looking at the code | |
| 12:38:43 | sean-k-mooney | i was using ubutnu 20.04 for my ceph testing | |
| 12:39:27 | sean-k-mooney | you can set CEPH_CONTAINERIZED=True in your local.conf | |
| 12:39:31 | sean-k-mooney | that might fix the issue for you | |
| 12:39:51 | Uggla | sean-k-mooney, it install this repo : https://download.ceph.com/rpm-pacific/el8/$basearch | |
| 12:40:08 | sean-k-mooney | yes | |
| 12:40:19 | sean-k-mooney | https://opendev.org/openstack/devstack-plugin-ceph/src/branch/master/devstack/lib/ceph#L31 | |
| 12:40:27 | sean-k-mooney | that is contolled by that ^ | |
| 12:41:48 | sean-k-mooney | Uggla: you should have hit this check https://opendev.org/openstack/devstack-plugin-ceph/src/branch/master/devstack/lib/ceph#L271-L28 | |
| 12:41:55 | sean-k-mooney | you ar enot running a supported os | |
| 12:43:00 | Uggla | sean-k-mooney, yep I had to add FORCE_CEPH_INSTALL=yes | |
| 12:43:16 | sean-k-mooney | ya that is why this is failing | |
| 12:43:23 | sean-k-mooney | so the repo config is contoled by https://opendev.org/openstack/devstack-plugin-ceph/src/branch/master/devstack/lib/ceph#L913-L943 | |
| 12:43:47 | bauzas | gibi: ack, thanks | |
| 12:44:14 | sean-k-mooney | Uggla: that is currently hardcoded to el8 | |
| 12:44:17 | sean-k-mooney | https://opendev.org/openstack/devstack-plugin-ceph/src/branch/master/devstack/lib/ceph#L1144 | |
| 12:44:18 | Uggla | sean-k-mooney, yep but in the repo there is no el9. | |
| 12:45:50 | sean-k-mooney | ya i see that | |
| 12:46:26 | Uggla | sean-k-mooney, CEPH_CONTAINERIZED=True, will install ceph in a container ? | |
| 12:46:32 | sean-k-mooney | yep | |
| 12:46:44 | sean-k-mooney | https://opendev.org/openstack/devstack-plugin-ceph/src/branch/master/devstack/lib/ceph#L157-L174 | |
| 12:47:02 | Uggla | sean-k-mooney, do you think it will work ? | |
| 12:47:05 | sean-k-mooney | so it will install ceph in a container and dinstall just the ceph client on the host form ceph_common | |
| 12:47:28 | Uggla | sean-k-mooney, I'm going to try... | |
| 12:47:34 | sean-k-mooney | i think it will be the only way to install ceph currently on centos 9 stream | |
| 12:48:17 | sean-k-mooney | Uggla: they are currently in the process or rewriting this plugin to use cephadm by the way | |
| 12:48:26 | sean-k-mooney | that might add centos 9 support as a result | |
| 12:49:00 | Uggla | sean-k-mooney, centos stream9 is currently a bit too more bleeding edge. :) | |
| 12:49:23 | sean-k-mooney | that is generally not a good thing | |
| 12:49:47 | sean-k-mooney | am so the way to work around this for centos 9 stream would be to use ceph form disto packages instead fo from upstream ceph | |
| 12:49:57 | sean-k-mooney | https://buildlogs.centos.org/centos/9-stream/storage/x86_64/ceph-quincy/ | |
| 12:50:40 | Uggla | sean-k-mooney, require docker of course any preference in between docker or podman ? | |
| 12:51:59 | sean-k-mooney | it will install it | |