| opendevreview | Merged openstack/kayobe stable/yoga: Fix forgotten hacluster regexp for image build https://review.opendev.org/c/openstack/kayobe/+/843323 | 07:07 |
|---|---|---|
| opendevreview | Verification of a change to openstack/kayobe stable/xena failed: Fix forgotten hacluster regexp for image build https://review.opendev.org/c/openstack/kayobe/+/843326 | 07:09 |
| opendevreview | Verification of a change to openstack/kayobe stable/wallaby failed: Fix forgotten hacluster regexp for image build https://review.opendev.org/c/openstack/kayobe/+/843327 | 08:33 |
| frickler | hrw: can't prove it yet but the timing of the latest qemu update seems to coincide with the arm job starting to fail https://metadata.ftp-master.debian.org/changelogs//main/q/qemu/qemu_5.2+dfsg-11+deb11u2_changelog | 08:56 |
| frickler | that's the only change I found in a diff with the last passing run | 08:57 |
| yoctozepto | hrw, frickler: may we run a different image? | 11:21 |
| opendevreview | Radosław Piliszek proposed openstack/kolla-ansible stable/ussuri: Cleanup zuul jobs a bit https://review.opendev.org/c/openstack/kolla-ansible/+/843639 | 11:49 |
| opendevreview | Radosław Piliszek proposed openstack/kolla stable/yoga: Use the new image naming scheme by default https://review.opendev.org/c/openstack/kolla/+/843644 | 12:00 |
| opendevreview | Radosław Piliszek proposed openstack/kolla stable/yoga: Use the new image naming scheme by default https://review.opendev.org/c/openstack/kolla/+/843644 | 12:01 |
| opendevreview | Radosław Piliszek proposed openstack/kolla stable/yoga: Fix Ubuntu image builds https://review.opendev.org/c/openstack/kolla/+/843644 | 12:02 |
| opendevreview | Radosław Piliszek proposed openstack/kolla-ansible master: [CI] Use the script module for running core tests https://review.opendev.org/c/openstack/kolla-ansible/+/843718 | 12:13 |
| frickler | yoctozepto: other than cirros you mean? that one has been unchanged for two years now afaict | 12:33 |
| opendevreview | Boxiang Zhu proposed openstack/kolla master: test https://review.opendev.org/c/openstack/kolla/+/843719 | 12:34 |
| opendevreview | Boxiang Zhu proposed openstack/kolla master: test https://review.opendev.org/c/openstack/kolla/+/843719 | 12:35 |
| frickler | new grafana panel is up, I already saw that I messed up centos jobs, let me know if you want more changes. I think I'll do kolla-asnsible and publish jobs for arm, too, or maybe rather a different page so that it doesn't get too full? or merge x86+arm into a single graph? | 12:41 |
| frickler | https://grafana.opendev.org/d/c0d59dad13/kolla-failure-rate?orgId=1 | 12:41 |
| frickler | yoctozepto: fyi, pretty old issue we found https://storyboard.openstack.org/#!/story/2006294 . so good that storyboard is super useful and folks are fixing things reported there all the time | 12:52 |
| yoctozepto | frickler: yeah, but maybe cirros is not handling qemu advancements well | 12:59 |
| yoctozepto | +2 on storyboard comment | 12:59 |
| frickler | yoctozepto: for a major version bump of qemu I could accept that, but not for a distro stable maintenance patch | 13:06 |
| frickler | although I must admit that I also see issues in my cirros test builds with the arm image. it works fine on bionic, fails on focal. but there the image does boot, it just doesn't find a network device https://review.opendev.org/c/cirros/cirros/+/842047 | 13:07 |
| yoctozepto | frickler: well, at qemu level I would expect that certain bug fixes may cause this anyway. I agree it's bad user experience though | 13:32 |
| frickler | I'd call it a regression. anyway, I'm going to hold another node and test this, maybe also try to boot a different image for comparison | 13:48 |
| opendevreview | Merged openstack/kolla-ansible stable/ussuri: Cleanup zuul jobs a bit https://review.opendev.org/c/openstack/kolla-ansible/+/843639 | 13:53 |
| *** hrww is now known as hrw | 15:12 | |
| yoctozepto | frickler: ack, good idea; in the meantime you might want to review https://review.opendev.org/c/openstack/kolla-ansible/+/843718 | 15:36 |
| yoctozepto | upgrades are tested well then | 15:36 |
| guesswhat | Can anyone help with connecting Octavia mngmt network ? I am kinda confused ( https://docs.openstack.org/kolla-ansible/latest/reference/networking/octavia.html ), I have only two networks, management ( eth0 ) and external/public ( eth1 ). Octavia uses eth0, but controller is not able to reach octavia worker... Any idea? | 15:46 |
| opendevreview | Radosław Piliszek proposed openstack/kolla-ansible master: Do not use keystone_admin_url et al https://review.opendev.org/c/openstack/kolla-ansible/+/843727 | 16:25 |
| opendevreview | Radosław Piliszek proposed openstack/kolla-ansible master: Fix a comment on openstack_interface https://review.opendev.org/c/openstack/kolla-ansible/+/843728 | 16:31 |
| opendevreview | Radosław Piliszek proposed openstack/kolla-ansible master: [WIP] Remove Keystone admin endpoint https://review.opendev.org/c/openstack/kolla-ansible/+/843730 | 16:40 |
| hrw | frickler: we use qemu 6.* from backports. | 17:41 |
| hrw | and it got bumped to 7.* on 5th May | 17:42 |
| frickler | my cirros issue was for arm, not aarch64, and I found the solution here https://bugs.launchpad.net/qemu/+bug/1790975 | 18:34 |
| frickler | hrw: ah, so major version bump that breaks this. qemu 6.2 was fine, 7.0 is broken (with cirros) | 20:07 |
| frickler | https://zuul.opendev.org/t/openstack/builds?job_name=kolla-ansible-debian-source-aarch64&project=openstack%2Fkolla-ansible&branch=master&result=SUCCESS&skip=0 | 20:08 |
| frickler | last pass https://zuul.opendev.org/t/openstack/build/98a34ecf5c074a14a8b0e3a881664745 | 20:08 |
| frickler | first fail https://zuul.opendev.org/t/openstack/build/2470ea9d4f37473da9c33c7ed471a5a7 | 20:08 |
| frickler | if you check the nova-compute log, you can see the virt-7.0 machine type appear in the latter | 20:09 |
| frickler | creating an instance with an ubuntu jammy cloud image works (at least it boots the kernel, fails some minutes later likely because of lack of memory) | 20:10 |
| guesswhat | Anyone is runing Octavia? Do I need to another flat network ( 1 for management, 1 for public network ) for Octavia management? Not sure how can I connect Octavia and Management network... Seems, its possible only via openvswitch bridging ( which is pain ), or if is flat, it would work out of the box, right ? | 22:00 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!