error: checksum verification failed

[proxwolfe]

Me too

There are already many threads on this topic and I tried to apply the solution to boot to debug and then import the zpool.

However, I am not sure how to get to the debug console: When I boot the proxmox installer (6.2), my monitor goes dark. If I press <return>, booting continues and I also get a picture again. This seems to be the normal installation mode. I am assuming that from the "dark" page, I can alternatively get into debug mode. So I boot again and this time, blindly, scroll down once with the arrow keys and then press return. The installer actually does boot into a debug console.

However, when I try to execute "zpool import ...", I get the error message: "/bin/sh: zpool: not found".

So, maybe this is the wrong debug console? How can I boot into a debug console that supports ZFS (keeping in mind that I can't see the first page after the installer loads)?

Or does it mean there is no zpool to import at all?

Many thanks

 

[Stoiko Ivanov]

I get the error message: "/bin/sh: zpool: not found".

that's a 'command not found' message

So, maybe this is the wrong debug console? How can I boot into a debug console that supports ZFS

exactly - the first debug console does not have the installer software yet available.
You should get to the debug-shell with zpool/zfs avaiable simply by pressing Ctrl+D in the shell you were
IOW: do what you did to get to "/bin/sh: zpool: not found" and then press Ctrl+D

I hope this explains it!

 

[proxwolfe]

Thanks for the quick reply!

Actually, I had tried that but nothing had happened (immediately) and so I rebooted (the server is not the fastest).

But knowing that something should work often helps to make it work.

So I booted again into the first debug shell and pressed Ctrl-D and, sure enough, it continued booting.

However, it has now been stuck at "[....] Waiting for /dev to the fully populated..." for over half an hour.

What do I do now?

 

[proxwolfe]

On second attempt no improvement: Over an hour stuck.

So, to the forum again. And I found also this problem is not new. However, it appears to occur with new installations and not, as in my case, during booting of the installer.

So it seems that the common cure, editing the grub config during boot, is not going to work for me. Or is there a way for me, in the first debug shell, to edit the grub config that will be used in the further boot process after I press Ctrl-D?

 

[proxwolfe]

Since I don't seem to be able to get to the correct debug shell, I thought I would try a different approach:

I added another disk to the system and installed a new instance of pve on it. It boots fine and it sees the other disks forming the zfs.

The next step would now be to import the old zfs pool into the new installation, right?

I found information on how to import a normal zpool into pve. However, the person said that this does not apply to an rpool. But my understanding is that I do have an rpool (because pve was installed directly on zfs).

My question is now: Do I have a zpool and an rpool - and can I simply import the zpool and ignore the rpool (because the new pve instance boots from another disk)? Or do I only have an rpool and need to import that one somehow?

Or won't importing be possible at all because of the checksum error anyway and I need to resolve this first? If so: How can I get to the correct debug shell, if my system does not boot beyond "waiting for /dev to be gully populated..."?

Thanks

 

[proxwolfe]

Update:

I assumed, for a start, that there is only the rpool and so I tried to import that using "zpool import -R /mnt rpool" but I got an error that the pool can't be imported because it was previously in use from another system. Fair enough.

So I tried to force it using "zpool import -f -R /mnt rpool" but I got an error that the directory "/mnt" was not empty.

So I tried to force it specifying a different mount point "zpool import -f -R /mnt/zfspool rpool" but I got an error that a pool with the name already exists. I am pretty sure it does not (my new installation is ext4).

So I tried to force import it giving it a new name "zpool import -f -R /mnt/zfspool rpool rpoolimp" but I got an error that no such pool is available.

But when I checked in the pve gui, the pool is now listed - I am not sure which of the commands above did the trick (I am guessing the first force import under /mnt because that's where it seems to be) but there it is.

But now what? How do I get to my VMs/CTs etc.?

Because my problem is: "zfs list" does show the rpool under "/mnt/rpool" but "ls /mnt" only shows "hostrun" and nothing else...

 

[proxwolfe]

Found something:

So the rpool is at /dev/rpool

But it only contains /data, nothing else (e.g. no /ROOT as listed by "zfs list").
In /data, there seem to be some VM disks (and CT disks?) but where is everything else, particularly the config files for the VMs/CTs?

 

[proxwolfe]

At this point, I am stuck.
Don't know what else to try.
Any ideas?

 

[proxwolfe]

Help?

 

[Stoiko Ivanov]

So I tried to force it specifying a different mount point "zpool import -f -R /mnt/zfspool rpool" but I got an error that a pool with the name already exists. I am pretty sure it does not (my new installation is ext4).

my guess is that the pool got imported but the ROOT dataset could not be mounted (because the alternative mountpoint was not empty)

could you try exporting the pool again and then importing it with the -R flag?

 

[proxwolfe]

Okay, so I exported the pool.

But instead of directly reimporting it again, I tried booting from it again. And it worked. The checksum error was gone.

I found out that booting from ZFS is not recommended for non-UEFI systems. So after booting successfully this time, I backed up my VMs/CTs and reinstalled pve without ZFS.

Thanks for your help!