| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2021-05-04 | |||
| 14:27:03 | sean-k-mooney | well its not a bug really | |
| 14:27:07 | bauzas | correct | |
| 14:27:07 | bauzas | correct | |
| 14:27:15 | bauzas | it's a non supported feature :) | |
| 14:27:15 | bauzas | it's a non supported feature :) | |
| 14:27:30 | sean-k-mooney | to disable cross az attach you have to align the cinder an nova az and have to froce a default az | |
| 14:27:30 | tobias-urdin | :D i'm not mad, i agree | |
| 14:27:30 | sean-k-mooney | to disable cross az attach you have to align the cinder an nova az and have to froce a default az | |
| 14:27:30 | tobias-urdin | :D i'm not mad, i agree | |
| 14:27:41 | bauzas | or a cinder bug about the cross_attach option, if you prefer | |
| 14:27:41 | bauzas | or a cinder bug about the cross_attach option, if you prefer | |
| 14:27:45 | sean-k-mooney | when users dont set one | |
| 14:27:46 | sean-k-mooney | when users dont set one | |
| 14:27:52 | bauzas | but, to fix the bug, you need a large spec | |
| 14:27:52 | bauzas | but, to fix the bug, you need a large spec | |
| 14:28:17 | bauzas | and honestly, I don't think nova needs to do the hardest and longest effort | |
| 14:28:17 | bauzas | and honestly, I don't think nova needs to do the hardest and longest effort | |
| 14:28:21 | tobias-urdin | we'll setting a default AZ doesn't help either, if you have older instances created before that was set i.e before there even was two AZ | |
| 14:28:21 | tobias-urdin | we'll setting a default AZ doesn't help either, if you have older instances created before that was set i.e before there even was two AZ | |
| 14:28:26 | tobias-urdin | the requestspec would still say az=none | |
| 14:28:26 | tobias-urdin | the requestspec would still say az=none | |
| 14:28:37 | tobias-urdin | it's messy | |
| 14:28:37 | tobias-urdin | it's messy | |
| 14:28:40 | bauzas | tobias-urdin: easy to fix, kill your pets :p | |
| 14:28:40 | bauzas | tobias-urdin: easy to fix, kill your pets :p | |
| 14:28:57 | sean-k-mooney | tobias-urdin: right greenfiled default az helps | |
| 14:28:57 | sean-k-mooney | tobias-urdin: right greenfiled default az helps | |
| 14:29:06 | sean-k-mooney | tobias-urdin: exising deployment not so much | |
| 14:29:06 | sean-k-mooney | tobias-urdin: exising deployment not so much | |
| 14:29:07 | bauzas | puppies are nice, but use a gun | |
| 14:29:07 | bauzas | puppies are nice, but use a gun | |
| 14:29:32 | bauzas | you want their leather | |
| 14:29:32 | bauzas | you want their leather | |
| 14:29:45 | tobias-urdin | bauzas: that would be the best thing, killing them but they're not mine :( | |
| 14:29:45 | tobias-urdin | bauzas: that would be the best thing, killing them but they're not mine :( | |
| 14:30:00 | tobias-urdin | i can send an angry email, that should work 2021 :pp | |
| 14:30:00 | tobias-urdin | i can send an angry email, that should work 2021 :pp | |
| 14:30:16 | bauzas | tobias-urdin: I know, I was unfortunately kidding | |
| 14:30:16 | bauzas | tobias-urdin: I know, I was unfortunately kidding | |
| 14:30:33 | bauzas | * bauzas would imagine himself dressed up like Cruella | |
| 14:31:07 | sean-k-mooney | those were dalmations | |
| 14:31:07 | tobias-urdin | * tobias-urdin dreams about deleting all affected instances | |
| 14:31:07 | sean-k-mooney | those were dalmations | |
| 14:31:15 | sean-k-mooney | anyway | |
| 14:31:15 | sean-k-mooney | anyway | |
| 14:31:23 | bauzas | tobias-urdin: in order to fix your existing instances, that's difficult | |
| 14:31:23 | bauzas | tobias-urdin: in order to fix your existing instances, that's difficult | |
| 14:31:31 | bauzas | as I said, the AZ value is within the RequestSpec | |
| 14:31:31 | bauzas | as I said, the AZ value is within the RequestSpec | |
| 14:31:40 | bauzas | and we persist the whole blob | |
| 14:31:40 | bauzas | and we persist the whole blob | |
| 14:32:17 | bauzas | you could hack tho with python scripts | |
| 14:32:17 | bauzas | you could hack tho with python scripts | |
| 14:32:28 | bauzas | and I'm pretty sure there are existing tools that make this for you | |
| 14:32:28 | bauzas | and I'm pretty sure there are existing tools that make this for you | |
| 14:33:00 | bauzas | * bauzas has seen a lof of operators asking how to modify the RequestSpec records | |
| 14:33:06 | sean-k-mooney | the way to "fix" this for your deployment would be two fold 1.) set a default az. 2.) write a scprit in python tha tloads the instance from the db and set the current az in its request spec | |
| 14:33:06 | sean-k-mooney | the way to "fix" this for your deployment would be two fold 1.) set a default az. 2.) write a scprit in python tha tloads the instance from the db and set the current az in its request spec | |
| 14:33:15 | bauzas | yup | |
| 14:33:26 | bauzas | but you can only have one default AZ | |
| 14:33:26 | bauzas | but you can only have one default AZ | |
| 14:33:38 | sean-k-mooney | there mihgt be something in the opts tools repo for this | |
| 14:33:38 | sean-k-mooney | there mihgt be something in the opts tools repo for this | |
| 14:34:00 | bauzas | this doesn't really help but you can 'load-balance' fakely by having different default AZs for each nova-api service | |
| 14:34:00 | bauzas | this doesn't really help but you can 'load-balance' fakely by having different default AZs for each nova-api service | |
| 14:34:10 | openstackgerrit | John Garbutt proposed openstack/nova-specs master: Re-propose Unified Limits in Nova https://review.opendev.org/c/openstack/nova-specs/+/789585 | |
| 14:34:10 | openstackgerrit | John Garbutt proposed openstack/nova-specs master: Re-propose Unified Limits in Nova https://review.opendev.org/c/openstack/nova-specs/+/789585 | |
| 14:34:16 | tobias-urdin | already have a default az, so newer instances actually works, it's just those old ones with the "legacy" requestspec data | |
| 14:34:16 | tobias-urdin | already have a default az, so newer instances actually works, it's just those old ones with the "legacy" requestspec data | |
| 14:34:30 | bauzas | tobias-urdin: script it then | |
| 14:34:30 | bauzas | tobias-urdin: script it then | |
| 14:34:50 | bauzas | tobias-urdin: you need to regenerate the RequestSpec object, then modify it, then persisting it again | |
| 14:34:50 | bauzas | tobias-urdin: you need to regenerate the RequestSpec object, then modify it, then persisting it again | |
| 14:35:30 | bauzas | that's pretty simple honestly, but you need to understand how oslo versioned objects work within nova | |
| 14:35:30 | bauzas | that's pretty simple honestly, but you need to understand how oslo versioned objects work within nova | |
| 14:35:53 | bauzas | that's a 3-line | |
| 14:35:53 | bauzas | that's a 3-line | |
| 14:36:05 | bauzas | <import the objects module> | |
| 14:36:05 | bauzas | <import the objects module> | |
| 14:36:22 | bauzas | <get the request_spec object from its UUID> | |
| 14:36:22 | bauzas | <get the request_spec object from its UUID> | |
| 14:36:30 | bauzas | <modify the object value> | |
| 14:36:30 | bauzas | <modify the object value> | |
| 14:36:34 | bauzas | <save the object> | |
| 14:36:34 | bauzas | <save the object> | |
| 14:37:06 | bauzas | (4 lines, actually) | |
| 14:37:06 | bauzas | (4 lines, actually) | |
| 14:39:03 | tobias-urdin | problem was that it's saved for neutron ports and cinder attachments as well right uggh | |
| 14:39:03 | tobias-urdin | problem was that it's saved for neutron ports and cinder attachments as well right uggh | |
| 14:40:13 | bauzas | * bauzas facepalms | |
| 14:40:49 | bauzas | KISS | |
| 14:40:50 | bauzas | KISS | |
| 14:41:14 | bauzas | (keep it simple) | |
| 14:41:14 | bauzas | (keep it simple) | |
| 14:41:44 | tobias-urdin | 15:51 < sean-k-mooney> for exampel we add the AZ to the neutron port in the device_owner filed i belive | |
| 14:41:44 | tobias-urdin | 15:51 < sean-k-mooney> for exampel we add the AZ to the neutron port in the device_owner filed i belive | |
| 14:41:47 | tobias-urdin | 15:52 < sean-k-mooney> its proably in the cinder attaments somewhere | |
| 14:41:47 | tobias-urdin | 15:52 < sean-k-mooney> its proably in the cinder attaments somewhere | |
| 14:42:10 | bauzas | tobias-urdin: you want to have the scheduler only supporting hosts from the AZ | |
| 14:42:10 | bauzas | tobias-urdin: you want to have the scheduler only supporting hosts from the AZ | |