Boot with ZFS failed with error 2

[eyebone]

Hello forum,

I am stuck at the boottime of a freshly installed FreeBSD System 9 on amd64 with zfs and geli.

During boottime I have to enter the geli passwords for the devices and the associated keys, this is working (I see the .eli devices in the mount prompt later). After successfully attach geli I get the error message:

Mounting from zfs:zpool failed with error 2.

I booted already several times into live fs and adjusted some of the properties like mountpoint to / and legacy, can mount on / off.

What else I might want to check?

Thanks,
Eyebone

 

[vermaden]

You need to 'regenerate' the /boot/zfs/zpool.cache file.

To do that, boot from FreeBSD live CD, then after attaching the GELI devices do:

# zpool import -f -o cachefile=/tmp/zpool.cache -o altroot=/mnt YourPoolName
# cp /tmp/zpool.cache /mnt/boot/zfs/zpool.cache
# zpool export YourPoolName
# zpool import YourPoolName

 

[eyebone]

You need to 'regenerate' the /boot/zfs/zpool.cache file.

To do that, boot from FreeBSD live CD, then after attaching the GELI devices do:

# zpool import -f -o cachefile=/tmp/zpool.cache -o altroot=/mnt YourPoolName
# cp /tmp/zpool.cache /mnt/boot/zfs/zpool.cache
# zpool export YourPoolName

Thanks! Your posting remembered me that I always give the cache the name of the pool and the boot loader is searching for zpool.cache and not for mypool.cache.

 

[vermaden]

Welcome.

 

[Markand]

I have the same problem, however the zpool.cache regeneration did not work for me, does the mount point property must be set to legacy on the pool?

 

[HarryE]

The pool must not be in "exported" state when trying to boot from it. Redo the

# zpool import -f -o cachefile=/tmp/zpool.cache -o altroot=/mnt YourPoolName
# cp /tmp/zpool.cache /mnt/boot/zfs/zpool.cache
# zpool export YourPoolName

then # zpool import YourPoolName and reboot

 

[Stochastix]

I first tried the @vermaden without reading to the end of the thread and it didn't work. The extra bit from @HarryE was required. Even though I had reset /boot/loader.conf to the new name of my root pool, I still had to enter it by hand at mountroute.

Strangely, I now see in /boot/loader.conf that

vfs.root.mountfrom="zfs:oldname/root"

This is neither the new name I see with zpool list (the same I had to give at the mountroute prompt). Is it going to boot next time without help? I don't know.

Would be nice if the critical @vermaden post could be edited with the extra bit in the fashion of StackExchange.

 

[vermaden]

@Stochastix,

Fixed.