Tom Parrott
Tom Parrott
> To stay backwards compatible if the flag isn't provided LXD won't wait for the resources to be ready. Or we default to a sensible timeout at the CLI level...
We effectively always need to check the status of networks and pools on the target member before triggering the restore otherwise we'll end up stopping instances on the source(s) and...
That would be my preference too.
> Furthermore a general bug is fixed in LXD which caused an instance's backup file not to be updated when updating the storage pool of its root volume. This action...
@roosterfish wonder if related to https://github.com/canonical/lxd/issues/15664
@roosterfish are any other instances being created concurrently?
@roosterfish can you see (by using an updated LXD using an improved wrapped error) if the issue is caused by this call? `limits.AllowInstanceCreation()` in `instancesPost`?
> By default, if no value is supplied, the pool size will be pulled from the global default pool size. Why dont we just leave the config setting unpopulated in...
Whilst I agree this is a bug, I am not entirely sure what we should return here for this endpoint. The problem is that for `physical` networks, the parent interface...
@mihalicyn please can you add what this fix resolves in terms of which ubuntu version we can now run as containers?