I've got a system using ZFS for its root filesystem, but unfortunately it's not using the GPT labels when accessing the block devices in the vdev. At first I didn't mind, but as I add disks to the system, things start to get really ugly, and now I have a mix of device node names (e.g. ada0p2), GPT IDs, and, if you consider all my pools, GPT labels as well.
I'd like to use GPT labels for everything, and the partitions that zroot resides on are already labeled. Having read this post, it seems that an export/import can fix things, but I wanted to clear up a few things in the special case of ZFS for root.
Code:
zpool status zroot
...
NAME STATE READ WRITE CKSUM
zroot ONLINE 0 0 0
raidz1-0 ONLINE 0 0 0
ada0p2 ONLINE 0 0 0
gptid/28cdc62c-b86f-11e1-a2be-003018a1d97f ONLINE 0 0 0
gptid/2d0c917e-b86f-11e1-a2be-003018a1d97f ONLINE 0 0 0
zpool status zbackup
...
NAME STATE READ WRITE CKSUM
zbackup ONLINE 0 0 0
gpt/backup ONLINE 0 0 0I'd like to use GPT labels for everything, and the partitions that zroot resides on are already labeled. Having read this post, it seems that an export/import can fix things, but I wanted to clear up a few things in the special case of ZFS for root.
- Is there anything particularly special about using ZFS root and GPT labels? I.e., is there any reason I shouldn't be able to use labels instead of device names/GPT IDs?
- What's the best way to perform such a procedure on a root pool? md root? Boot from another medium?
- Any issues caused by having the ZFS mountpoint being "legacy"? Will I need to change it temporarily in order to export/import?
- Any other pitfalls I should know about?