Persistent State Partition Questions

Are there some examples of what state consists of?
Is it a lot of data written?
How do you recover from Persistent State Partition drive death?

Can the Persistent State be mirrored: zpool, btrfs, lvm, etc?
Can the Persistent State be stored with MinIO distributed within a cluster?
Can the Persistent State be stored with Longhorn? (When its available)

In the cloud-config.yml, can you specify the disks in the various ways listed:
/dev/sda
/dev/by-label/rancher_state
/dev/by-uuid/0a3407de-014b-458b-b5c1-848e92a327a3
/dev/by-path/…

The docs page does show dev: LABEL=RANCHER_STATE.
I was curious if you can do by-path specifically.

the documentation states that you can use
devices and labels and uuid
http://rancher.com/docs/os/v1.1/en/storage/state-partition/

do by path is possible (same as /dev/sda1 etc)

Good to know that by-path works as expected. I assume you could do UUID or LABEL the long way as well, but the LABEL= and UUID= were provided as a shortcut.

I also found:

It seems that only EXT4 is supported at the moment. So no zpools, mirrors, etc. (EDIT: for the persistent state partition).

http://rancher.com/docs/os/v1.1/en/storage/using-zfs/

The linked guide is definitely a good guide for using ZFS with RancherOS. But it is not quite the same as RancherOS on ZFS root; which I am not sure is possible.

As far as my previous comment. I think ros install will always use EXT4 if you turn on auto-format, or have not already manually partitioned. Otherwise, you can manually format any extlinux supported file system? (FAT12/16/32, NTFS, ext2/3/4, Btrfs, XFS, UFS/FFS). Then use fstype: auto, XFS, btrfs, etc.

Here is something I found for using mdraid: