Repository navigation
Various bugfixes - #3352
Various bugfixes#3352
Conversation
Skip OVN setup when network.ovn.northbound_connection is empty so nodes without OVS installed can apply the config change after the cluster unsets OVN. During cluster join, drop OVN config keys from the trigger changes map since the explicit setup calls already handle them with error tolerance, preventing a join from failing on nodes that lack OVS. Closes lxc#3351 Signed-off-by: Stéphane Graber <stgraber@stgraber.org>
GetResources only cares about top-level statistics and so shouldn't be triggering a full cache run. This was making cluster state retrieval extremely slow. Signed-off-by: Stéphane Graber <stgraber@stgraber.org> Sponsored-by: https://webdock.io
Signed-off-by: Stéphane Graber <stgraber@stgraber.org>
Signed-off-by: Stéphane Graber <stgraber@stgraber.org>
|
LINSTOR looks unhappy, I’ll rerun the checks |
|
The error has been ignored in a few PRs now. It first appeared in #3338, which suggests it’s an upstream change. sigh https://github.com/LINBIT/linstor-server/blob/master/CHANGELOG.md#1333---2026-05-11 Any idea what’s causing a shrink anyway @stgraber? |
|
@bensmrs The standalone case appears to be an actual container shrink test, so that's now failing given they removed support for it. The cluster case is less obvious but may also be effectively the same thing. The pool's default volume size would be 10GiB. If the container's profile instead sets something smaller, then a shrink would happen on creation which would similarly fail. So I think the errors are correct from the Incus side, it's the tests that will need updating and probably a note added to the documentation. |
lxc/lxc#4635 Signed-off-by: Stéphane Graber <stgraber@stgraber.org>
|
Ok, I’ll have a look |
No description provided.