| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2021-04-21 | |||
| 09:46:25 | openstackgerrit | Wenping Song proposed openstack/nova-specs master: Add no user token when get Cyborg client https://review.opendev.org/c/openstack/nova-specs/+/787178 | |
| 11:04:30 | openstackgerrit | Stephen Finucane proposed openstack/nova master: Remove references to 'sys.version_info' https://review.opendev.org/c/openstack/nova/+/787348 | |
| 11:04:30 | openstackgerrit | Stephen Finucane proposed openstack/nova master: Drop support for SQLite < 3.7 https://review.opendev.org/c/openstack/nova/+/787349 | |
| 11:04:30 | openstackgerrit | Stephen Finucane proposed openstack/nova master: Remove references to 'sys.version_info' https://review.opendev.org/c/openstack/nova/+/787348 | |
| 11:04:30 | openstackgerrit | Stephen Finucane proposed openstack/nova master: Drop support for SQLite < 3.7 https://review.opendev.org/c/openstack/nova/+/787349 | |
| 12:11:48 | sean-k-mooney | do we specify which db coalation type should be used? | |
| 12:11:48 | sean-k-mooney | do we specify which db coalation type should be used? | |
| 12:12:02 | sean-k-mooney | i dont see it in the db models | |
| 12:12:02 | sean-k-mooney | i dont see it in the db models | |
| 12:13:53 | sean-k-mooney | i locally restored a copy of a customer db and the dump does not have a coalation type recorded and i realise the queries i am doing will have different results based on if i use utf8_general_ci which is the mysql default vs utf8_bin | |
| 12:13:53 | sean-k-mooney | i locally restored a copy of a customer db and the dump does not have a coalation type recorded and i realise the queries i am doing will have different results based on if i use utf8_general_ci which is the mysql default vs utf8_bin | |
| 12:17:25 | gibi | sean-k-mooney: what I see it that the FlavorExtraSpec api db model has a specific arg {'mysql_collate': 'utf8_bin'}, | |
| 12:17:25 | gibi | sean-k-mooney: what I see it that the FlavorExtraSpec api db model has a specific arg {'mysql_collate': 'utf8_bin'}, | |
| 12:21:33 | sean-k-mooney | for just that one table which we dont use any more :) | |
| 12:21:33 | sean-k-mooney | for just that one table which we dont use any more :) | |
| 12:21:51 | sean-k-mooney | i thinke the rest use whatever you have set as your default type | |
| 12:21:51 | sean-k-mooney | i thinke the rest use whatever you have set as your default type | |
| 12:22:39 | sean-k-mooney | we have a customer which for reasons ended up changing the nova hosts value form lowercase to upper case | |
| 12:22:39 | sean-k-mooney | we have a customer which for reasons ended up changing the nova hosts value form lowercase to upper case | |
| 12:23:00 | sean-k-mooney | so they have 2 compute node recored and had 2 compute service records but they deleted one | |
| 12:23:00 | sean-k-mooney | so they have 2 compute node recored and had 2 compute service records but they deleted one | |
| 12:23:18 | sean-k-mooney | but they have instnace with both the upercase and lower case host value | |
| 12:23:18 | sean-k-mooney | but they have instnace with both the upercase and lower case host value | |
| 12:23:54 | sean-k-mooney | and with teh utf8_general_ci coalation type instance.host will match either the upercase or lowercase compute node recored | |
| 12:23:54 | sean-k-mooney | and with teh utf8_general_ci coalation type instance.host will match either the upercase or lowercase compute node recored | |
| 12:24:24 | sean-k-mooney | SELECT cn.uuid , cn.host, i.host, i.deleted from nova.compute_nodes cn join nova.instances i on cn.host = i.host WHERE i.deleted = 0; | |
| 12:24:24 | sean-k-mooney | SELECT cn.uuid , cn.host, i.host, i.deleted from nova.compute_nodes cn join nova.instances i on cn.host = i.host WHERE i.deleted = 0; | |
| 12:24:45 | sean-k-mooney | i.e. if i do that cn.host and i.host can have different cases | |
| 12:24:45 | sean-k-mooney | i.e. if i do that cn.host and i.host can have different cases | |
| 12:25:31 | sean-k-mooney | i dont think we actully intended that to be how that works so i think we really should be using utf8_bin for all the tables | |
| 12:25:31 | sean-k-mooney | i dont think we actully intended that to be how that works so i think we really should be using utf8_bin for all the tables | |
| 12:26:08 | sean-k-mooney | that might be what we say in the doc and or in the installer tools by setting the db coalation type to utf8_bin | |
| 12:26:08 | sean-k-mooney | that might be what we say in the doc and or in the installer tools by setting the db coalation type to utf8_bin | |
| 12:26:20 | sean-k-mooney | but i just noticed it does not seam to be enforced by our schema | |
| 12:26:20 | sean-k-mooney | but i just noticed it does not seam to be enforced by our schema | |
| 12:26:27 | sean-k-mooney | so i was not sure if this was intentional or not | |
| 12:26:27 | sean-k-mooney | so i was not sure if this was intentional or not | |
| 12:35:10 | sean-k-mooney | gibi: the actual issue the customer is hitting is that the conductor/api and compute agent disagree on what the name of the compute node queue should be | |
| 12:35:10 | sean-k-mooney | gibi: the actual issue the customer is hitting is that the conductor/api and compute agent disagree on what the name of the compute node queue should be | |
| 12:36:00 | sean-k-mooney | rabbitmq cares about the case but we are not enforcing that and depending on when the vms were booted some have the uppercases name and other have the lower case name | |
| 12:36:00 | sean-k-mooney | rabbitmq cares about the case but we are not enforcing that and depending on when the vms were booted some have the uppercases name and other have the lower case name | |
| 12:37:43 | gibi | I see. I'm not against fixing it in our schema but then we need to provide an upgrade script that normalize the casing too | |
| 12:37:43 | gibi | I see. I'm not against fixing it in our schema but then we need to provide an upgrade script that normalize the casing too | |
| 12:43:16 | sean-k-mooney | well im not sure if we can fix it without breaking things | |
| 12:43:16 | sean-k-mooney | well im not sure if we can fix it without breaking things | |
| 12:43:36 | sean-k-mooney | at least i dont think we can do it without requiring a data migration to fix the things that will break | |
| 12:43:36 | sean-k-mooney | at least i dont think we can do it without requiring a data migration to fix the things that will break | |
| 12:43:54 | sean-k-mooney | and likely a nova-status check for some edgecases | |
| 12:43:54 | sean-k-mooney | and likely a nova-status check for some edgecases | |
| 12:44:25 | gibi | true, it is not just a schema change but also some data normalization issue | |
| 12:44:25 | gibi | true, it is not just a schema change but also some data normalization issue | |
| 12:44:46 | sean-k-mooney | gibi: right now im trying to figure out why it was semi operational in this state and how to resovle it for this specific customer | |
| 12:44:46 | sean-k-mooney | gibi: right now im trying to figure out why it was semi operational in this state and how to resovle it for this specific customer | |
| 12:45:13 | sean-k-mooney | i suspect its because they change the nova.conf host value several times | |
| 12:45:13 | sean-k-mooney | i suspect its because they change the nova.conf host value several times | |
| 12:46:35 | sean-k-mooney | i think ill need to get them to use the same case on all host and then etiehr modify the service host value or the instance.host | |
| 12:46:35 | sean-k-mooney | i think ill need to get them to use the same case on all host and then etiehr modify the service host value or the instance.host | |
| 12:47:04 | sean-k-mooney | unless we can make rabbit not care about case but really thats a hack | |
| 12:47:04 | sean-k-mooney | unless we can make rabbit not care about case but really thats a hack | |
| 12:48:49 | sean-k-mooney | i guess one way to fix it would be to normalise just the rabbitmq names | |
| 12:48:49 | sean-k-mooney | i guess one way to fix it would be to normalise just the rabbitmq names | |
| 12:49:02 | sean-k-mooney | for example make them all lowercase | |
| 12:49:02 | sean-k-mooney | for example make them all lowercase | |
| 12:49:43 | sean-k-mooney | we could do that with a workaround config option and backport that maybe | |
| 12:49:43 | sean-k-mooney | we could do that with a workaround config option and backport that maybe | |
| 12:59:54 | openstackgerrit | Merged openstack/nova master: Replace blind retry with libvirt event waiting in detach https://review.opendev.org/c/openstack/nova/+/770246 | |
| 12:59:54 | openstackgerrit | Merged openstack/nova master: Replace blind retry with libvirt event waiting in detach https://review.opendev.org/c/openstack/nova/+/770246 | |
| 13:00:20 | gibi | sean-k-mooney: yeah that could work | |
| 13:00:21 | gibi | sean-k-mooney: yeah that could work | |
| 13:01:14 | openstackgerrit | Merged openstack/nova master: Move the guest.get_disk test to test_guest https://review.opendev.org/c/openstack/nova/+/777151 | |
| 13:01:14 | openstackgerrit | Merged openstack/nova master: Move the guest.get_disk test to test_guest https://review.opendev.org/c/openstack/nova/+/777151 | |
| 13:01:54 | openstackgerrit | Merged openstack/nova master: Enable mypy on libvirt/guest.py https://review.opendev.org/c/openstack/nova/+/777155 | |
| 13:01:54 | openstackgerrit | Merged openstack/nova master: Enable mypy on libvirt/guest.py https://review.opendev.org/c/openstack/nova/+/777155 | |
| 13:02:36 | openstackgerrit | Merged openstack/nova master: Follow up type hints for a634103 https://review.opendev.org/c/openstack/nova/+/777159 | |
| 13:02:36 | openstackgerrit | Merged openstack/nova master: Follow up type hints for a634103 https://review.opendev.org/c/openstack/nova/+/777159 | |
| 13:06:45 | sean-k-mooney | gibi: are we only doing 3 hours each day at the ptg instead of 4 | |
| 13:06:45 | sean-k-mooney | gibi: are we only doing 3 hours each day at the ptg instead of 4 | |
| 13:06:52 | gibi | yes | |
| 13:07:01 | sean-k-mooney | oh ok why so short? | |
| 13:07:01 | sean-k-mooney | oh ok why so short? | |
| 13:07:20 | gibi | 14:00 UTC is still pretty early in west coast | |
| 13:07:20 | gibi | 14:00 UTC is still pretty early in west coast | |
| 13:07:26 | gibi | US | |
| 13:07:26 | gibi | US | |
| 13:07:34 | sean-k-mooney | ya it is | |
| 13:07:34 | sean-k-mooney | ya it is | |
| 13:08:31 | sean-k-mooney | i dont know i guess i just felt 3 days was kind of short to begin with but on the otherhand i guess i cant to the in person whiteborad sessions with peopel on topic remotely so maybe not | |
| 13:08:31 | sean-k-mooney | i dont know i guess i just felt 3 days was kind of short to begin with but on the otherhand i guess i cant to the in person whiteborad sessions with peopel on topic remotely so maybe not | |
| 13:08:55 | sean-k-mooney | if we go back to in person ptgs next cycle it would be nice to have more time again | |
| 13:08:55 | sean-k-mooney | if we go back to in person ptgs next cycle it would be nice to have more time again | |
| 13:09:23 | gibi | if it feels that we are not progressing then I can add extra time | |
| 13:09:23 | gibi | if it feels that we are not progressing then I can add extra time | |
| 13:09:42 | sean-k-mooney | ack | |
| 13:09:58 | sean-k-mooney | * sean-k-mooney only realised it was shorter because i tought i was already late :) | |
| 13:10:23 | gibi | I felt that using 13:00 UTC would be unfair for west coast, so I assume that EU can take a bit more time even after 17:)) UTC | |
| 13:10:23 | gibi | I felt that using 13:00 UTC would be unfair for west coast, so I assume that EU can take a bit more time even after 17:)) UTC | |
| 13:10:25 | sean-k-mooney | * sean-k-mooney now i can be late because i tought i was early | |
| 13:10:43 | gibi | I mean after 17:00 | |
| 13:10:43 | gibi | I mean after 17:00 | |
| 13:10:54 | sean-k-mooney | gibi: ya no that is fine i was happy to go later on the poll | |