ZFS Pool suddenly can not be imported/mounted

tuaris

tuaris

This is a virtual machine, and it's not had any issues/problems since it was last booted and properly shutdown. I have a pool that will now not mount, even though I was using it a day prior:

Code: 
 zpool import Home
cannot import 'Home': I/O error
        Destroy and re-create the pool from
        a backup source.

The pool itself is ONLINE when I check it with zpool import:
Code: 
zpool import
   pool: jails
     id: 7320448665102965750
  state: ONLINE
status: Some supported features are not enabled on the pool.
        (Note that they may be intentionally disabled if the
        'compatibility' property is set.)
 action: The pool can be imported using its name or numeric identifier, though
        some features will not be available without an explicit 'zpool upgrade'.
 config:

        jails          ONLINE
          gpt/jails01  ONLINE

   pool: Home
     id: 17578580742360951213
  state: ONLINE
 action: The pool can be imported using its name or numeric identifier.
 config:

        Home        ONLINE
          da1p1     ONLINE

I am currently using ZFS on Root, and zpool status shows it's there are no issues with that:

Code: 
 zpool status
  pool: FreeBSD
 state: ONLINE
  scan: scrub repaired 0B in 00:09:55 with 0 errors on Mon Jul 27 13:44:23 2020
config:

        NAME        STATE     READ WRITE CKSUM
        FreeBSD     ONLINE       0     0     0
          da0p3     ONLINE       0     0     0

errors: No known data errors
 
I think for the foreseeable future I am going to hold off on using ZFS inside virtual machines on the desktop/workstation space. I've lost way too much data over the years due to different types of failures. This is probably the 6th time I've lost some meaningful (not so critical) data. ZFS may not be suitable for this use case. There's a pile of dead bodies stacking up... VMDK's that I can't bring myself to delete even though the ZFS pools on them are not importable. :'‑(
 
As another anecdote, never happened to me with dozens of ZFS VMs some of which have seen quite a lot of beating (during FreeBSD and ZFS development) over 10+ years.
 
Out of curiousity, OP what is the host running? Is it FreeBSD or something else? Not sure it makes a difference, but it may.
 
I do understand your frustration but I can also say that I've never had any issues with ZFS on any of the hypervisor I use (VMware Fusion/Workstation, qemu and VirtualBox).

tuaris said:
VMDK's that I can't bring myself to delete even though the ZFS pools on them are not importable
Click to expand...

You can delete vmdk regardless of what's on it. On running VM disk is in use. Depending on the type of the disk you may be able to detach and remove it online. If not, you can always remove the vmdk when VM is offline. If not you have most likely something else going on that is propagating to your VM.

You didn't provide any information about your setup, how VM is configured and the layout of the disks. It's important to know what is the storage backend for those vmdk disks too.
 
It's VMware Workstation on Windows 10 as the host. I have been using VMware for decades and almost never had issues with data loss except for when I use ZFS. The backend storage for this specific case is a NVMe, but I've also used HDD and SDD's. I just use a single disk for ZFS in the guest and no duplication.

To clarify... it's not that I can't delete the broken VMDK's. It's that I don't want to, in the hopes that one day I may find a way to recover that data. Even though I have been able to rebuild or recouperate parts of it from ocasional backups or extra copies.
 
Any chance you have vmdk that doesn't have sensitive, publicly shareable data?
Could you attach the VM config ?

The only thing that comes to my mind is some weird I/O caching settings on host side.

On Fusion (a bit less on Workstation) I did abuse VMs a lot many times ending up crashing them. Never had any issues importing pools. I know, doesn't help you much.
 
You must log in or register to reply here.
Back
Top