noonedeadpunk | logan-: nice to see you, hope things are good :) | 07:57 |
---|---|---|
opendevreview | Niklas Schwarz proposed openstack/openstack-ansible-rabbitmq_server master: Add ability to set environment variable RABBITMQ_USE_LONGNAMES https://review.opendev.org/c/openstack/openstack-ansible-rabbitmq_server/+/901003 | 08:20 |
opendevreview | Niklas Schwarz proposed openstack/openstack-ansible-rabbitmq_server master: Add ability to set environment variable RABBITMQ_USE_LONGNAMES https://review.opendev.org/c/openstack/openstack-ansible-rabbitmq_server/+/901003 | 08:32 |
jrosser | ^ i am wondering if this needs the cert contents to reflect the FQDN or not | 10:14 |
noonedeadpunk | ++ good point | 10:16 |
jrosser | i would hope that the infra job will show this | 10:37 |
jrosser | but needs a careful look | 10:37 |
nixbuilder | noonedeadpunk: I'm sorry... but the patch did not seem to work :-( Here is the log... it seems like the same error: https://paste.opendev.org/show/bUc4skUIDxtNKl5S0bkN/ | 10:45 |
nixbuilder | noonedeadpunk: And here is my glance-api.conf file which includes the changes that jrosser indicated in the link he sent me https://paste.openstack.org/show/bxBTJ9F9CUlAY8aw2fh0/ | 10:49 |
noonedeadpunk | nixbuilder: so how does the /etc/glance/rootwrap.conf look like? | 10:52 |
nixbuilder | noonedeadpunk: https://paste.opendev.org/show/bJIyibNh5V4aIGNN4nNs/ | 10:53 |
noonedeadpunk | hm. that looks like if patch wasn't applied actually... | 10:54 |
noonedeadpunk | In CI it looks as expected and you should have the line like that: https://zuul.opendev.org/t/openstack/build/773b3beadec145c5b0d935a35fac9758/log/logs/etc/host/glance/rootwrap.conf.txt#20 | 10:55 |
nixbuilder | noonedeadpunk: Hmmmm.... | 10:57 |
jrosser | noonedeadpunk: i see your question in the magnum meeting about how to build the control plane with limited connectivity | 10:58 |
jrosser | i made this as a quick hack https://github.com/jrosser/make-controlplane-registry | 10:58 |
noonedeadpunk | oh, that's sweet. I haven't started looking into all CAPI topic as that's planned only for the next year, but trying to check on things to be aware... | 10:59 |
jrosser | same approach would be able to host anything in a private registry | 10:59 |
jrosser | just needs the upstream things listing, currently it's a bit hardcoded but totally works in a test deployment here | 11:00 |
jrosser | and i put an example set of user_variables in there which would be needed for the vexxhost driver | 11:01 |
noonedeadpunk | btw, do you use poetry widely? | 11:02 |
jrosser | nope, this is just total lift/shift from something mnaser already wrote to make a registry for the user clusters | 11:03 |
noonedeadpunk | as I'm slightly... sceptical/uncertain about it right now. Without any good reason, but simple constraints feel easier to read.. dunno | 11:03 |
noonedeadpunk | and some folks pushing hard for it right now... | 11:04 |
noonedeadpunk | gotcha | 11:04 |
jrosser | anyway, airgapped totally works | 11:05 |
nixbuilder | noonedeadpunk: OK... I am not sure what happened with the patch, but when I edited the exec_dirs line in the /etc/glance/rootwrap.conf file, the image upload worked! | 11:15 |
nixbuilder | noonedeadpunk: BTW... I double checked the file and the patch is in there... so I don't know :-( | 11:16 |
noonedeadpunk | huh | 12:08 |
noonedeadpunk | that;s weird.... | 12:09 |
noonedeadpunk | nixbuilder: are you doing source install or distro one? | 12:11 |
nixbuilder | noonedeadpunk: The default... I think that is source. Although I could have applied the patch to the wrong files??? I applied them to the files in '/etc/ansible/roles/os_glance'. | 12:18 |
noonedeadpunk | nah, that is correct location | 12:19 |
nixbuilder | noonedeadpunk: Here is the results of the particular files after patching... looks good put I may have missed something: https://paste.opendev.org/show/bE0ljtxXz0t0i4Gx10Fq/ | 12:30 |
noonedeadpunk | yeah, that looks pretty much good... | 12:37 |
noonedeadpunk | would be interesting to see tasks output then.... | 12:37 |
opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-os_glance master: Make glance_additional_stores work nicely with multi-store https://review.opendev.org/c/openstack/openstack-ansible-os_glance/+/901034 | 13:41 |
opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-os_glance master: Make glance_additional_stores work nicely with multi-store https://review.opendev.org/c/openstack/openstack-ansible-os_glance/+/901034 | 13:43 |
opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-os_glance master: Make glance_additional_stores work nicely with multi-store https://review.opendev.org/c/openstack/openstack-ansible-os_glance/+/901034 | 13:49 |
nixbuilder | noonedeadpunk: There was one more anomaly I noticed... after installation glance-api would not start because 'enabled_backends = cinder:cinder,cinder:cinder' in /etc/glance/glance-api.conf. The glance-api complained that there was a duplicate backed and would not start. | 13:50 |
nixbuilder | noonedeadpunk: Of course editing out the duplicate fixed the issue. | 13:51 |
noonedeadpunk | yeah, that's because you have cinder both in glance_default_store and glance_additional_stores | 13:51 |
noonedeadpunk | you should drop it from additional_stores if it's in default | 13:52 |
nixbuilder | noonedeadpunk: Gotcha | 13:52 |
spatel | I have 32 core compute and I want to make it double like 64 core in that case what ratio I should be adding for nova ? 2.0 ? | 14:03 |
noonedeadpunk | um, it won't make compute 64 cores, but it would allow to have up to 64 vCPUs for VMs | 14:26 |
noonedeadpunk | so yes | 14:26 |
noonedeadpunk | if that's what you mean | 14:26 |
jrosser | if its hyperthreaded it might be sort of doing 64 anyway? | 14:26 |
jrosser | i forget if cores or threads are counted | 14:27 |
mgariepy | in my case the my limiting factor is either local nvme or ram. | 14:27 |
mgariepy | unless placement is messing things up ;) haha | 14:27 |
noonedeadpunk | they are | 14:28 |
opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: Remove glance_available_stores logic from group_vars https://review.opendev.org/c/openstack/openstack-ansible/+/901041 | 14:37 |
opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-os_glance master: Make glance_additional_stores work nicely with multi-store https://review.opendev.org/c/openstack/openstack-ansible-os_glance/+/901034 | 14:38 |
opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-os_glance master: Make glance_additional_stores work nicely with multi-store https://review.opendev.org/c/openstack/openstack-ansible-os_glance/+/901034 | 14:47 |
opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-os_glance master: Fix cinder store configuration https://review.opendev.org/c/openstack/openstack-ansible-os_glance/+/901046 | 15:09 |
noonedeadpunk | ugh, glance indeed appeared to be a rabbit hole.... | 15:10 |
spatel | noonedeadpunk thanks.. I want to put more vm so trying to adjust ratio | 16:13 |
noonedeadpunk | well, enabling HT can be another way indeed | 16:16 |
admin1 | i accidentally mapped an IP to a wrong node .. i deleted it from inventory and compute service .. and gave the correct node to the IP .. but when i provision it, it still comes up wiht the old hostname | 17:20 |
admin1 | so where in the tables is a uuid/ip to hostname mapped for nova compute agent list | 17:21 |
mgariepy | hmm live-migration fail on same model hosts with 2023.1 ? | 19:54 |
mgariepy | does it ring a bell to someone ? | 19:54 |
mgariepy | Migration pre-check error: Unacceptable CPU info: CPU doesn't have compatibility. | 19:55 |
mgariepy | :? | 19:55 |
mgariepy | does feel like : but .. why ?https://bugs.launchpad.net/nova/+bug/1898715 | 19:57 |
mgariepy | https://bugs.launchpad.net/nova/+bug/2039803 | 20:16 |
mgariepy | fun. | 20:17 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!