incusd/instance/qemu: Cleanup volume eject/detach logic #2330
Conversation
This removes some duplicate logic, avoids failures when volumes come from profiles and adds special-handling for our internal volumes. Signed-off-by: Stéphane Graber <stgraber@stgraber.org> Sponsored-by: https://webdock.io
|
@bensmrs any issue with this? It simplifies the logic a bit and fixes an issue where RHEL based systems would only boot successfully once as after that |
| @@ -5935,8 +5910,27 @@ func (d *qemu) detachDisk(name string) error { | |||
| return err | |||
| } | |||
|
|
|||
| // Check if it's a special device (we don't store detached state on those). | |||
| if slices.Contains([]string{"agent:config", "cloud-init:config"}, config["source"]) { | |||
There was a problem hiding this comment.
To me, this would only trigger if the agent is added explicitly in Incus config. It avoids a round of DB transactions leading to an ignored error, though.
There was a problem hiding this comment.
incus config device add MY-INSTANCE agent disk source=agent:config is what you have to do for all RHEL based images. All of those suddenly failed to start the agent with the addition of the attached property.
There was a problem hiding this comment.
Oh ok, I forgot how RHEL was painful to handle :)
| } | ||
|
|
||
| // Find the disk device. | ||
| _, ok = d.localDevices[diskName] |
There was a problem hiding this comment.
I’m not sure how I didn’t think about using localDevices in the PR.
|
Yeah, I did a quick test both with the special drives and with a regular one, behaved as expected. |
This removes some duplicate logic, avoids failures when volumes come from profiles and adds special-handling for our internal volumes.
Sponsored-by: https://webdock.io