| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-os_masakari master: Add tag to enable targeting of post-install config elements only https://review.opendev.org/c/openstack/openstack-ansible-os_masakari/+/919690 | 07:02 |
|---|---|---|
| noonedeadpunk | what do we need to land before branching actually? | 07:11 |
| noonedeadpunk | I see like ironic and trove patches for qmanager kinda | 07:11 |
| noonedeadpunk | aha, rabbitmq flags | 07:14 |
| andrewbonney | I think those and ideally the k8s related ones assuming we can get through the capi job which jrosser is still investigating intermittent fails for | 07:14 |
| jrosser | noonedeadpunk: I had one query on the trove patch too | 07:16 |
| noonedeadpunk | yep, answered it | 07:16 |
| noonedeadpunk | guestagent is kinda amphora thing for trove | 07:16 |
| jrosser | great will look again shortly | 07:16 |
| noonedeadpunk | so it runs on a vm | 07:16 |
| jrosser | aah ok | 07:17 |
| noonedeadpunk | unless I'm very wrong about something | 07:17 |
| jrosser | but the service end also needs to connect to its different mq? | 07:17 |
| noonedeadpunk | ok, that is a good point actually | 07:18 |
| jrosser | that’s what I was thinking that for trove there might be two totally separate mq? and we use the same shm | 07:18 |
| noonedeadpunk | yeah | 07:18 |
| noonedeadpunk | but.... how we did that | 07:19 |
| noonedeadpunk | huh | 07:22 |
| noonedeadpunk | how that worked lol | 07:22 |
| noonedeadpunk | but it apparently did as I had trove back then deployed and working | 07:24 |
| noonedeadpunk | I'm gonna deploy trove in some poc very shortly, so will extra test this different rabbit part.... | 07:43 |
| noonedeadpunk | so yeah, it could be worth to have a different hostname for guestagent just in case.... | 07:47 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-os_trove master: Define unique hostname for QManager https://review.opendev.org/c/openstack/openstack-ansible-os_trove/+/920455 | 07:50 |
| jrosser | if it turns out to be wrong in some way it's only a config override needed to fix it until we can patch a bug | 07:52 |
| jrosser | is this made more complicated by the need for a 2nd rabbit instance being optional | 07:53 |
| * jrosser not really knowing if the shm/qmanager is per queue or per connection to rabbit | 07:54 | |
| noonedeadpunk | I guess it's just a need of 2nd rabbit which makes it complicated | 07:54 |
| noonedeadpunk | so it's kinda per process thread | 07:54 |
| noonedeadpunk | so - connection | 07:54 |
| jrosser | so if we have two things (services) on the same host using the same queue they still need unique shm | 07:55 |
| jrosser | which i guess is where this all started | 07:55 |
| noonedeadpunk | yeah | 07:58 |
| andrewbonney | Looks like all CI is failing cloning keystone at the moment | 08:18 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: Enable oslomsg_rabbit_queue_manager by default https://review.opendev.org/c/openstack/openstack-ansible/+/920715 | 08:19 |
| noonedeadpunk | doh | 08:19 |
| noonedeadpunk | but kinda how.... that should be coming from zuul? | 08:20 |
| noonedeadpunk | git clone --filter=blob:none --quiet file:///openstack/src/opendev.org/openstack/keystone failing, yeah | 08:22 |
| jrosser | noonedeadpunk: its git safe dirs? fatal: detected dubious ownership in repository at '/openstack/src/opendev.org/openstack/keystone/.git | 10:54 |
| jrosser | To add an exception for this directory, call: git config --global --add safe.directory /openstack/src/opendev.org/openstack/keystone/.git | 10:55 |
| jrosser | i wonder why this starts only now | 10:55 |
| noonedeadpunk | huh | 10:56 |
| noonedeadpunk | I thought we were adding smth globally | 10:56 |
| * noonedeadpunk still kinda away at event | 10:56 | |
| jrosser | it's odd as it doesnt happen in the jammy upgrade job | 10:56 |
| noonedeadpunk | and kinda tough to reproduce locally | 10:57 |
| noonedeadpunk | so we have that I guess now: https://opendev.org/openstack/openstack-ansible/commit/efa153ea159af13ce5539e423371f2642c9592ef | 10:58 |
| noonedeadpunk | maybe it needs to be more global now.... | 10:58 |
| jrosser | that would cover the metal jobs to make that more global | 10:59 |
| jrosser | but lxc is whole other thing https://review.opendev.org/c/openstack/openstack-ansible/+/920715?tab=change-view-tab-header-zuul-results-summary | 10:59 |
| jrosser | here we go https://github.com/git/git/commit/f4aa8c8bb11dae6e769cd930565173808cbb69c8 | 11:04 |
| noonedeadpunk | pf | 11:07 |
| jrosser | i think there was a point release of git for jammy yesterday | 11:07 |
| noonedeadpunk | that would explain it all | 11:08 |
| noonedeadpunk | I wonder, if it woud make sense to do smth with zuul job of placing repos actually... | 11:08 |
| jrosser | still for lxc when we do not build wheels that setting needs to be in all the service containers | 11:09 |
| jrosser | its a client setting rather than at the source of the repo | 11:09 |
| noonedeadpunk | yeah, true | 11:10 |
| noonedeadpunk | for containers its indeed worse kinda | 11:10 |
| noonedeadpunk | I was just thinking that nodepool VM is still managed by zuul, so... | 11:11 |
| noonedeadpunk | but yeah, won't help us | 11:11 |
| jrosser | we can add a task in openstack_hosts perhaps | 11:12 |
| jrosser | and some variable | 11:12 |
| noonedeadpunk | I was thinking if we can kinda pass a client file with alredy existing vars... | 11:13 |
| noonedeadpunk | As I assume we do bind map zuul prepared repos to containers? | 11:14 |
| noonedeadpunk | sorry, I'm kinda brain-farting today | 11:14 |
| jrosser | that right they are bind mount in /openstack to the same place in containers as on the host, i think | 11:16 |
| jrosser | so that the repo setup is identical between metal and lxc jobs | 11:16 |
| jrosser | i can try to make a patch later | 11:19 |
| noonedeadpunk | ok, would be very much appreciated, as I'm on internal meetup for next 2 days, so really hard to do smth... | 11:23 |
| frickler | thx for the pointer to the git update, seems kolla is affected, too | 12:39 |
| jrosser | frickler: did you end up doing safe.directory = * ? | 15:22 |
| jrosser | i dont thing you can do safe.directory = /all/my/repos/* | 15:23 |
| opendevreview | Jonathan Rosser proposed openstack/openstack-ansible-openstack_hosts master: Allow openstack_hosts role to configure git safe.directory entries https://review.opendev.org/c/openstack/openstack-ansible-openstack_hosts/+/920757 | 15:37 |
| opendevreview | Jonathan Rosser proposed openstack/openstack-ansible master: Configure all directories to be safe for git clone in CI https://review.opendev.org/c/openstack/openstack-ansible/+/920758 | 15:42 |
| opendevreview | Jonathan Rosser proposed openstack/openstack-ansible-openstack_hosts master: Do not mark all directories as safe for git clone by default https://review.opendev.org/c/openstack/openstack-ansible-openstack_hosts/+/920759 | 15:45 |
| opendevreview | Jonathan Rosser proposed openstack/openstack-ansible-openstack_hosts master: Allow openstack_hosts role to configure git safe.directory entries https://review.opendev.org/c/openstack/openstack-ansible-openstack_hosts/+/920757 | 16:02 |
| opendevreview | Jonathan Rosser proposed openstack/openstack-ansible-openstack_hosts master: Do not mark all directories as safe for git clone by default https://review.opendev.org/c/openstack/openstack-ansible-openstack_hosts/+/920759 | 16:03 |
| opendevreview | Jonathan Rosser proposed openstack/openstack-ansible-openstack_hosts master: Do not mark all directories as safe for git clone by default https://review.opendev.org/c/openstack/openstack-ansible-openstack_hosts/+/920759 | 16:03 |
| opendevreview | Jonathan Rosser proposed openstack/openstack-ansible-openstack_hosts master: Do not mark all directories as safe for git clone by default https://review.opendev.org/c/openstack/openstack-ansible-openstack_hosts/+/920759 | 16:26 |
| opendevreview | Jonathan Rosser proposed openstack/openstack-ansible-openstack_hosts master: Allow openstack_hosts role to configure git safe.directory entries https://review.opendev.org/c/openstack/openstack-ansible-openstack_hosts/+/920757 | 16:50 |
| opendevreview | Jonathan Rosser proposed openstack/openstack-ansible-openstack_hosts master: Do not mark all directories as safe for git clone by default https://review.opendev.org/c/openstack/openstack-ansible-openstack_hosts/+/920759 | 16:50 |
| frickler | jrosser: didn't look into a solution yet | 17:36 |
| jrosser | i ended up with brute force safe.directory = "*" | 17:36 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!