ZFS Livelist Bug -- When will TrueNAS Merge Patch?

[yottabit]

Hi Everyone,

Several months ago I ran into a kernel panic reboot loop after cloning a zvol. I engaged directly with the OpenZFS team, along with several others experiencing the same issue, and they were able to isolate and fix the problem with this submitted patch.

I was able to boot an Ubuntu live image and mount the zpool successfully. Later, I was able to boot TrueNAS CORE successfully without experiencing the kernel panic.

Now fast forward to this week; I attempted to delete the zvol clone that originally caused this problem, and now I'm back in the kernel panic reboot loop.

Does anyone know when this OpenZFS patch will merge into TrueNAS CORE? I believe I'm running 12.0-U1 or 12.0-U2 at the moment.

Thank you!

 

[sretalla]

Well, 12.0-U5 merged ZFS 2.0.5

/_archive/corereleasenotes/12.0/12.0u5/

I have no idea what version your patch equates to.

 

[yottabit]

According to the github Projects pane, it looks like they merged the patch into 2.1, but not yet 2.0 (awaiting triage). :-/

 

[Kris Moore]

We're looking into back-porting to 2.0.6 now, thanks for the heads up.

 

[yottabit]

Thank you, Kris! Any guesstimates on how long it should take to release a CORE build with the patch? I won't hold you to it of course, just wanting to get a rough idea.

It looks like SCALE merged OpenZFS 2.1.0.

So if I migrate from CORE to SCALE, I should be able to regain access to my zpool. But of course that means running a beta in prod, refactoring my jails to containers, and migrating my VMs from bhyve to KVM.

Thanks again!

 

[Kris Moore]

Thank you, Kris! Any guesstimates on how long it should take to release a CORE build with the patch? I won't hold you to it of course, just wanting to get a rough idea.

It looks like SCALE merged OpenZFS 2.1.0.

So if I migrate from CORE to SCALE, I should be able to regain access to my zpool. But of course that means running a beta in prod, refactoring my jails to containers, and migrating my VMs from bhyve to KVM.

Thanks again!

What, you aren't ready to YOLO it?

Probably 12.0-U7 is my guess, not sure if it'll make the cut-off for U6 yet. We'll see.

 

[freqlabs]

The commit applies cleanly to zfs-2.0.6-staging so I've proposed it be included in that here: https://github.com/openzfs/zfs/pull/12545#issuecomment-916980609

I don't expect this to be merged in time for 12.0-U6 but it should easily make U7.

 

[yottabit]

Can you confirm that SCALE BETA.1 includes the livelist patch?

Code:

[   16.095825] ZFS: Loaded module v2.1.0-35_g9dc7eb109 (DEBUG mode), ZFS pool version 5000, ZFS filesystem version 5

I think this version should have the patch, but I'm still experiencing long, blocking txg. But at least it isn't causing a kernel panic as on CORE.

 

[freqlabs]

Can you confirm that SCALE BETA.1 includes the livelist patch?

Code:

[   16.095825] ZFS: Loaded module v2.1.0-35_g9dc7eb109 (DEBUG mode), ZFS pool version 5000, ZFS filesystem version 5

I think this version should have the patch, but I'm still experiencing long, blocking txg. But at least it isn't causing a kernel panic as on CORE.

Yes it's present in that module.

 

[yottabit]

Yes it's present in that module.

Cheers. I will mention this to OpenZFS. It seems the patch may not be adequate, or there is a secondary issue that they have not addressed.