| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2021-05-20 | |||
| 10:42:17 | sean-k-mooney | but we can discuss that in the implemantion | |
| 10:42:57 | sean-k-mooney | artom: the only time it would be allowable but ill advised to change the hostname today is if there are no instnaces on the host | |
| 10:42:57 | sean-k-mooney | artom: the only time it would be allowable but ill advised to change the hostname today is if there are no instnaces on the host | |
| 10:43:46 | artom | sean-k-mooney, oh, I agree with you in principle | |
| 10:43:46 | artom | sean-k-mooney, oh, I agree with you in principle | |
| 10:43:51 | artom | But you know how operators be ;) | |
| 10:43:51 | artom | But you know how operators be ;) | |
| 10:45:05 | openstackgerrit | norman shen proposed openstack/nova master: Saving security group to info_cache https://review.opendev.org/c/openstack/nova/+/786348 | |
| 10:45:06 | admin0 | mine was an oversight.. unintentional | |
| 10:45:06 | admin0 | mine was an oversight.. unintentional | |
| 10:45:45 | admin0 | midway we decided to not have .openstack.local but the playbooks updated every hostname ( including existing ones) | |
| 10:45:45 | admin0 | midway we decided to not have .openstack.local but the playbooks updated every hostname ( including existing ones) | |
| 10:46:17 | admin0 | what creates a new queue in rabbitmq? is it the compute node ? | |
| 10:46:17 | admin0 | what creates a new queue in rabbitmq? is it the compute node ? | |
| 10:46:42 | artom | admin0, they usually are, and this is what sean-k-mooney is trying to prevent. But I can easily imagine someone saying "no no, we know what we're doing, we're totally going to rename this host" | |
| 10:46:42 | artom | admin0, they usually are, and this is what sean-k-mooney is trying to prevent. But I can easily imagine someone saying "no no, we know what we're doing, we're totally going to rename this host" | |
| 10:47:14 | admin0 | i also think its a lack of documentation or warning . i don't recall reading this otherwise we would have put it in the policy | |
| 10:47:15 | admin0 | i also think its a lack of documentation or warning . i don't recall reading this otherwise we would have put it in the policy | |
| 10:47:28 | artom | Fair point. I guess we thought it was implicit? | |
| 10:47:28 | artom | Fair point. I guess we thought it was implicit? | |
| 10:48:28 | admin0 | you guys know the code and for you all.. its implicit .. we ( operators) read the deployment guide and make policies based on the guides .. so this point got overlooked | |
| 10:48:28 | admin0 | you guys know the code and for you all.. its implicit .. we ( operators) read the deployment guide and make policies based on the guides .. so this point got overlooked | |
| 10:49:13 | artom | "you guys know the code" Should... should we tell them? ;) | |
| 10:49:13 | artom | "you guys know the code" Should... should we tell them? ;) | |
| 10:49:23 | admin0 | :D | |
| 10:49:23 | admin0 | :D | |
| 10:49:51 | sean-k-mooney | artom: operators tend to get annoy wehn we tell them we explictly dont support this so your on your own to fix your mess | |
| 10:49:52 | sean-k-mooney | artom: operators tend to get annoy wehn we tell them we explictly dont support this so your on your own to fix your mess | |
| 10:51:15 | sean-k-mooney | admin0: we should have documenation for this as far as i know we dissuced adding it a few times | |
| 10:51:15 | sean-k-mooney | admin0: we should have documenation for this as far as i know we dissuced adding it a few times | |
| 10:55:22 | sean-k-mooney | admin0: i still cant finx the warning we said shold be added like 2 years ago :( | |
| 10:55:22 | sean-k-mooney | admin0: i still cant finx the warning we said shold be added like 2 years ago :( | |
| 10:56:42 | sean-k-mooney | admin0: not to dismay you but this is one of the harder thing to fix to the point that the only supporte way to actully change the host name is generally to scale in and redeploy | |
| 10:56:42 | sean-k-mooney | admin0: not to dismay you but this is one of the harder thing to fix to the point that the only supporte way to actully change the host name is generally to scale in and redeploy | |
| 11:09:36 | admin0 | sean-k-mooney, if i have a way to migraete ceph backed instances off this hypervisor to another one, i would happilfy redeploy | |
| 11:09:36 | admin0 | sean-k-mooney, if i have a way to migraete ceph backed instances off this hypervisor to another one, i would happilfy redeploy | |
| 11:09:55 | admin0 | i don't understand nova :) .. https://gist.githubusercontent.com/a1git/9d4f3115529a53793703ef08bdfd752b/raw/ceec98a6145170143146d9dbd276156c5a52278f/gistfile1.txt | |
| 11:09:55 | admin0 | i don't understand nova :) .. https://gist.githubusercontent.com/a1git/9d4f3115529a53793703ef08bdfd752b/raw/ceec98a6145170143146d9dbd276156c5a52278f/gistfile1.txt | |
| 11:10:04 | admin0 | i changed all to h3 .. then it still adds as h3. | |
| 11:10:04 | admin0 | i changed all to h3 .. then it still adds as h3. | |
| 11:11:17 | sean-k-mooney | is h3. in the instance.host | |
| 11:11:17 | sean-k-mooney | is h3. in the instance.host | |
| 11:12:18 | sean-k-mooney | by the way libvirt also caches info | |
| 11:12:18 | sean-k-mooney | by the way libvirt also caches info | |
| 11:13:11 | sean-k-mooney | so i think you will need to restart libvirt when you make these change before you restart the compute agent | |
| 11:13:11 | sean-k-mooney | so i think you will need to restart libvirt when you make these change before you restart the compute agent | |
| 11:13:57 | sean-k-mooney | have you restarted the nova-compute agent after the hostname change | |
| 11:13:58 | sean-k-mooney | have you restarted the nova-compute agent after the hostname change | |
| 11:15:40 | admin0 | i have restarted the nova-compute agent only | |
| 11:15:41 | admin0 | i have restarted the nova-compute agent only | |
| 11:15:44 | admin0 | and not libvirt | |
| 11:15:44 | admin0 | and not libvirt | |
| 11:15:53 | admin0 | will restarting libvirt cause a downtime to the guests ? | |
| 11:15:53 | admin0 | will restarting libvirt cause a downtime to the guests ? | |
| 11:15:58 | sean-k-mooney | no | |
| 11:16:11 | sean-k-mooney | there lifetime is governe by that of the qemu process that is runnign them | |
| 11:16:11 | sean-k-mooney | there lifetime is governe by that of the qemu process that is runnign them | |
| 11:16:36 | admin0 | virsh hostname returns h3 and not h3. | |
| 11:16:36 | admin0 | virsh hostname returns h3 and not h3. | |
| 11:16:48 | sean-k-mooney | libvirt and the nova comptue agent can be restarted without affectign the vms | |
| 11:16:48 | sean-k-mooney | libvirt and the nova comptue agent can be restarted without affectign the vms | |
| 11:17:05 | sean-k-mooney | well thats what its returing to you not nessialry what its returing to nova | |
| 11:17:05 | sean-k-mooney | well thats what its returing to you not nessialry what its returing to nova | |
| 11:17:22 | admin0 | "is h3. in the instance.host" -- where do I see this | |
| 11:17:22 | admin0 | "is h3. in the instance.host" -- where do I see this | |
| 11:17:25 | sean-k-mooney | for reasons they can be different | |
| 11:17:25 | sean-k-mooney | for reasons they can be different | |
| 11:17:41 | sean-k-mooney | in the cell db | |
| 11:17:41 | sean-k-mooney | in the cell db | |
| 11:18:16 | sean-k-mooney | select * form nova_cell1.instance where host == 'h3.'; | |
| 11:18:17 | sean-k-mooney | select * form nova_cell1.instance where host == 'h3.'; | |
| 11:18:35 | sean-k-mooney | you can also get this in an instnace show | |
| 11:18:35 | sean-k-mooney | you can also get this in an instnace show | |
| 11:18:37 | admin0 | [nova_cell0]> select * from compute_nodes | |
| 11:18:37 | admin0 | [nova_cell0]> select * from compute_nodes | |
| 11:18:40 | admin0 | i only have cell0 | |
| 11:18:40 | admin0 | i only have cell0 | |
| 11:18:42 | sean-k-mooney | as an admin | |
| 11:18:42 | sean-k-mooney | as an admin | |
| 11:18:45 | admin0 | this is an old ( rocky) platform | |
| 11:18:45 | admin0 | this is an old ( rocky) platform | |
| 11:18:55 | sean-k-mooney | admin0: do you also have a db call just nova | |
| 11:18:55 | sean-k-mooney | admin0: do you also have a db call just nova | |
| 11:19:02 | sean-k-mooney | you should have 3 nova dbs | |
| 11:19:02 | sean-k-mooney | you should have 3 nova dbs | |
| 11:19:20 | sean-k-mooney | nova_cell0 nova_api and a third one? | |
| 11:19:20 | sean-k-mooney | nova_cell0 nova_api and a third one? | |
| 11:20:08 | admin0 | https://gist.github.com/a1git/9bc0c00aaacb3e8757b40e678a3be0fc are the dbs i ahve | |
| 11:20:08 | admin0 | https://gist.github.com/a1git/9bc0c00aaacb3e8757b40e678a3be0fc are the dbs i ahve | |
| 11:20:28 | sean-k-mooney | yep so the nova db is the cell1 db | |
| 11:20:28 | sean-k-mooney | yep so the nova db is the cell1 db | |
| 11:23:18 | admin0 | select count(*) from nova.instances where host='h3.' = > 0 ; select count(*) from nova.instances where host='h3' => 71 | |
| 11:23:18 | admin0 | select count(*) from nova.instances where host='h3.' = > 0 ; select count(*) from nova.instances where host='h3' => 71 | |
| 11:23:55 | admin0 | nova_cell0.instances has zero rows | |
| 11:23:56 | sean-k-mooney | ok | |
| 11:23:56 | admin0 | nova_cell0.instances has zero rows | |
| 11:23:56 | sean-k-mooney | ok | |
| 11:24:28 | sean-k-mooney | so you have restart both libvirt and nova compute and confirm the hypervior hostname in the in the db is h3 | |
| 11:24:28 | sean-k-mooney | so you have restart both libvirt and nova compute and confirm the hypervior hostname in the in the db is h3 | |
| 11:25:31 | openstackgerrit | Stephen Finucane proposed openstack/nova master: Add 'hw:vif_multiqueue_enabled' flavor extra spec https://review.opendev.org/c/openstack/nova/+/792356 | |