kubelet won't start on LXD with btrfs rootfs

[thedude66]

I tried deploying charmed kubernetes with juju on a debian bullseye server running snapd and LXD. Everything went well until it was time to wait for the worker nodes to register with the master.
Every worker node's kubelet was crash looping with an error along the lines of "Failed to start ContainerManager" err="failed to get rootfs info: failed to get device for dir \"/var/lib/kubelet\": could not find device with major: 0, minor: 269 in cached partitions map". A bit of googling suggests this is an upstream problem related to btrfs support.
It appears this can (at least sort-of temporarily) be worked around by setting passing --feature-gates="LocalStorageCapacityIsolation=false" to the kubelet command line arguments. I performed this cahnge manually and my workers came up. Unfortunately there is no way to pass feature gates using a juju overlay or anything of the sort.

I suggest either adding this feature gate by default if the charm is being installed on LXD with a btrfs root or adding a way to pass feature gates with an overlay.