*** mhen_ is now known as mhen | 01:32 | |
opendevreview | Takashi Kajinami proposed openstack/devstack master: Use yaml format policy file https://review.opendev.org/c/openstack/devstack/+/957066 | 05:18 |
---|---|---|
*** haleyb|out is now known as haleyb | 12:32 | |
opendevreview | Takashi Kajinami proposed openstack/tempest master: Drop description about allow_overlapping_ips option https://review.opendev.org/c/openstack/tempest/+/957146 | 14:35 |
opendevreview | Takashi Kajinami proposed openstack/devstack master: Fix ownership of keyring file https://review.opendev.org/c/openstack/devstack/+/957153 | 16:04 |
gmaan | sean-k-mooney: tosky : Seeing no objection on ML, I have added both of you in devstack core group, you will be able to +2 on devstack changes -https://review.opendev.org/admin/groups/412309ac864036d5323b496bb5e42cd000d7829d,members | 16:47 |
sean-k-mooney | gmaan: thanks i noticed a few minutes ago | 16:48 |
sean-k-mooney | i was just discussign this on the nova changell https://paste.opendev.org/show/blWPtsQswhTSAlgU6Ozb/ the luks issue we were seeing with qemu seam to not happen anymore but there is a much less frequent failure happing on debian | 16:49 |
sean-k-mooney | https://tinyurl.com/3phwkcyk that the opensearch hits | 16:49 |
sean-k-mooney | if i remove the nova filter its still on 377 hits in the past 30 days or the n days of logs we ahve aviable which is much lower then what we saw before | 16:50 |
sean-k-mooney | im not sure if this is a cidner bug with the "throttle_cmd" | 16:51 |
frickler | sean-k-mooney: is there a bug report for that issue? seems all the failure happen on raxflex, so possible correlated to newer CPUs that are available there | 16:54 |
tosky | thank you | 16:54 |
sean-k-mooney | frickler: no i just noticed it a few minutes ago. and ya i was wonderign the same. initally i was wonderign if it was realted to the xen issue we had with cpus not beign reported correctly but that is runing kvm correct? | 16:55 |
sean-k-mooney | frickler: it looks like that message is coming form qemu-img | 16:57 |
sean-k-mooney | https://bugzilla.redhat.com/show_bug.cgi?id=2336437 | 16:58 |
sean-k-mooney | apprently its fixed in rawhid i wonder in which release | 16:58 |
sean-k-mooney | qemu-9.2.0-3.fc42 | 16:59 |
sean-k-mooney | thats unfortunet debian 12 has 10.0.2 in the backprot repo but 7.2 in main it seams | 17:01 |
sean-k-mooney | ubuntu 24.04 is still on 8.2.2 apprently so that would not help | 17:01 |
sean-k-mooney | although im looking at https://repology.org/project/qemu/versions so that might not be up to date | 17:02 |
sean-k-mooney | rocky 9 is on 9.1.0 so also too old same as stream 9 | 17:03 |
sean-k-mooney | i assume we are not mirring the debian backport repo? | 17:04 |
tkajinam | https://opendev.org/opendev/system-config/src/branch/master/playbooks/roles/reprepro/files/debian/config/distributions#L46-L57 | 17:06 |
tkajinam | I guess we do | 17:06 |
clarkb | sean-k-mooney: I think we do mirror backports | 17:07 |
sean-k-mooney | ok then we could consider enabling that to work around this or monitor it to see when/if it gets resovled | 17:07 |
clarkb | and yes raxflex is kvm not xen. | 17:07 |
sean-k-mooney | i assume we are just not enableing the backprot repo in the jobs/image build | 17:08 |
sean-k-mooney | looks like kolla is using them already today | 17:09 |
sean-k-mooney | https://opendev.org/openstack/kolla/src/branch/master/docker/base/sources.list.debian | 17:10 |
sean-k-mooney | but we are not doing that in the devstack jobs. ill push up a test patch to see what happens if i add that | 17:10 |
sean-k-mooney | https://opendev.org/zuul/zuul-jobs/src/branch/master/roles/configure-mirrors/templates/apt/etc/apt/sources.list.j2#L5 | 17:12 |
sean-k-mooney | looks like we can do that but we dont by default | 17:12 |
sean-k-mooney | hum no apprently we do... https://opendev.org/zuul/zuul-jobs/src/branch/master/roles/configure-mirrors/defaults/main.yaml#L14 | 17:17 |
sean-k-mooney | but we are defintly not using it for qemu https://storage.gra.cloud.ovh.net/v1/AUTH_dcaab5e32b234d56b626f72581e3644c/zuul_opendev_logs_ec4/openstack/ec46632352504e2b925489e676dbb982/controller/logs/libvirt/libvirt/qemu/instance-00000001_log.txt | 17:18 |
sean-k-mooney | qemu version: 7.2.17Debian 1:7.2 | 17:18 |
clarkb | sean-k-mooney: you have to explicitly request to install from backports by default | 17:19 |
sean-k-mooney | oh | 17:19 |
clarkb | enabling the repo doesn't mean every packages is installed from there. It allows you to apt-get install qemu/backports or something like that | 17:19 |
sean-k-mooney | ok so that actully better | 17:19 |
sean-k-mooney | as we can opt in for just qemu/libvit | 17:20 |
clarkb | I forget the specific syntax but you need to be explicit about installing from backports along those lines | 17:20 |
sean-k-mooney | ack | 17:20 |
sean-k-mooney | `sudo apt install <package-name>/<codename>-backports` or `sudo apt install -t <codename>-backports <package-name>` acording to gemiini | 17:22 |
sean-k-mooney | the -t approch soudn more correct | 17:23 |
sean-k-mooney | which means either pluming in a new env var to contole this to here https://github.com/openstack/devstack/blob/7cc18281964e96424e6aad5f8c993725b80e32bb/functions-common#L1432-L1440 and potitally adding a new branch here https://github.com/openstack/devstack/blob/master/lib/nova_plugins/functions-libvirt#L71 | 17:25 |
sean-k-mooney | or to just test if this works i can drop in local.sh with a apt install comamdn with a pre-palybook jsut to see if that works | 17:26 |
clarkb | or just update the package name in the package file list to test it | 17:27 |
sean-k-mooney | i think ill do the quick hack for now just to get a ci run and then look at this again later in the week | 17:27 |
clarkb | I think the package files listing may have support for release specific packages. If it does then you could use that as the solution | 17:27 |
clarkb | just annotate each qemu/foo-backports entry with a note that it belongs to the foo release | 17:27 |
sean-k-mooney | they do but we dont install qemu that way | 17:27 |
clarkb | oh | 17:28 |
clarkb | maybe we should :) | 17:28 |
sean-k-mooney | althogh i proably can pre isntall it and then it wont care | 17:28 |
sean-k-mooney | its currently done here https://github.com/openstack/devstack/blob/master/lib/nova_plugins/functions-libvirt#L58-L112 | 17:28 |
sean-k-mooney | so we can avoid installing it in jobs that done use it | 17:28 |
sean-k-mooney | at least i think taht is the orginal logic | 17:29 |
clarkb | could move qemu proper out of that function and let the function handle the libvirt-python install only | 17:29 |
clarkb | that would work as long as we install qemu via the package file list first I think (since that is a dep of the python source build of libvirt-python) | 17:29 |
sean-k-mooney | your suggesting modifying https://github.com/openstack/devstack/blob/master/files/debs/nova#L16 | 17:29 |
sean-k-mooney | that does allwo distro filtering but im not sure where this is actully used | 17:31 |
sean-k-mooney | for example https://github.com/openstack/devstack/blob/master/files/rpms/n-cpu#L3 | 17:31 |
clarkb | ya so you'd do something like qemu/bookworm-backports # debian:bookwork or whatever the matcher syntax is | 17:32 |
clarkb | then you don't need any special code to handle it | 17:32 |
sean-k-mooney | ya the quistion i have is when is that package list used, is it before or after the existign function | 17:33 |
clarkb | ah I think you can answer that from a devstack log | 17:33 |
sean-k-mooney | https://github.com/openstack/devstack/blob/7cc18281964e96424e6aad5f8c993725b80e32bb/functions-common#L1294C10-L1294C22 | 17:33 |
clarkb | just look for when dosfstools is installed vs when libvirt-python is installed | 17:33 |
sean-k-mooney | i think its form here https://github.com/openstack/devstack/blob/7cc18281964e96424e6aad5f8c993725b80e32bb/tools/install_prereqs.sh#L64 | 17:35 |
sean-k-mooney | although in the neutron code we alsu use it directly https://github.com/openstack/devstack/blob/7cc18281964e96424e6aad5f8c993725b80e32bb/lib/neutron_plugins/ovs_base#L76 | 17:35 |
sean-k-mooney | so here https://github.com/openstack/devstack/blob/7cc18281964e96424e6aad5f8c993725b80e32bb/stack.sh#L804 | 17:36 |
sean-k-mooney | ok before wasting a bunch of ci time ill add "creating a debian 12" dev vm to my list for tomorrow and try and play around with this locally | 17:38 |
sean-k-mooney | frickler: by the way i deployed your debian 13 patch this morning and wa able to run tempest-full with no errors | 17:39 |
opendevreview | Douglas Mendizábal proposed openstack/devstack master: Fix iniset to escape backslash characters https://review.opendev.org/c/openstack/devstack/+/956914 | 19:10 |
opendevreview | Douglas Mendizábal proposed openstack/devstack master: Fix iniset to escape backslash characters https://review.opendev.org/c/openstack/devstack/+/956914 | 19:39 |
opendevreview | Douglas Mendizábal proposed openstack/devstack master: Fix iniset to escape backslash characters https://review.opendev.org/c/openstack/devstack/+/956914 | 19:41 |
opendevreview | Merged openstack/devstack master: Bubble up image download failures https://review.opendev.org/c/openstack/devstack/+/950273 | 20:12 |
opendevreview | Merged openstack/devstack master: Support CentOS Stream 10 https://review.opendev.org/c/openstack/devstack/+/937251 | 20:44 |
opendevreview | Merged openstack/devstack master: Remove temporary fix for problems with CSS styles https://review.opendev.org/c/openstack/devstack/+/949066 | 20:44 |
opendevreview | Clark Boylan proposed openstack/tempest master: DNM Test devstack and tempest with Ansible 11 in Zuul https://review.opendev.org/c/openstack/tempest/+/957186 | 21:49 |
opendevreview | Clark Boylan proposed openstack/tempest master: DNM Test devstack and tempest with Ansible 11 in Zuul https://review.opendev.org/c/openstack/tempest/+/957186 | 22:40 |
Generated by irclog2html.py 4.0.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!