Support volume filesystem expansion on volume expansion

[thorn3r]

Feature Request

Support volume filesystem expansion when a Block Storage Volume is resized

---

Summary

When a Block Storage Volume is expanded the filesystem is not resized. This results in the Volume being larger and costing more money, but the usable storage remains the same. To make use of the expanded storage space, a user must shell into the node it's attached to and resize the filesystem on the Volume.

This was mentioned in the #linode channel on the Kubernetes slackspace, where a user mentioned a similar issue was raised on another provider, who implemented this with piraeusdatastore/linstor-csi#70

[igorbrigadir]

I came up against this recently too - having to SSH into a node to run a command seems really unnecessary for expanding a volume.

[wbh1]

Info dumping here for myself (or whoever gets around to working on this):

• How AWS EBS driver does it
• Relevant k8s.io/mount-utils pkg
• Where I think our code needs updated

[rl0nergan]

Started working on this, but discovered that even when manually rescanning the drive via echo 1 > /sys/block/$devName/device/rescan the new device size isn't detected so filesystem can't be resized. Only solution I've found to detect the new device size is to reboot the node or detach/re-attach the volume. More details on this limitation can be found here, but until this is supported online filesystem expansion doesn't appear to be possible