I have a bunch of different zpools attached to my main machine. I’m having terribly annoying issues with kernel panics in one of datasets.
Most of the time it’s a
Blockquote
kernel: VERIFY3(0 == dmu_bonus_hold (os, object, dl, &dl->dl_dbuf)) failed (0 == 52)
Blockquote
Message from syslogd@fedora at Sep 1 10:31:13 kernel: PANIC at dsl_deadlist.c:308: dsl_deadlist_open()
Which is currently an open issue on GitHub, related to changing the block size and the compression types.
I’ve been trying to delete the dataset, snapshots, files in there, etc and EVERY time it panics before I can actually get it to complete. This dataset has 10tb of shows in it, the data isn’t irreplaceable, I have dumped a few TB of the contents out to other drives.
This pool (media-pool), it’s 9 drive 6tb z3 pool, it’s also in the middle of trying to resolver a disk. It gets to ~98% to every time and then the kernel panics and I have to restart it.
I have the ZFS pause scrub sysctl set at boot to stop it from resilvering over and over. I need to figure out how to fix this dataset before letting it go.
Every time it hangs I have to ‘reboot -f’. Currently running fedora server on it with the 6.1 LTS kernel and zfs2.x.
I tried switching to zfs-fuse (0.7) in the fedora repo, wouldn’t import.
I’ve also tried truenas core, similar results.
Every time I try deleting anything from that dataset it all seems to come back. I have all my zfs pools mounted sync=always to help with the panics but it’s not helping much.
I’ve tried just removing the dataset normally, ie zpool destroy -Rf dataset, the cursor just hangs.
I’ve tried removing the snapshots from the dataset, it panics before finishing.
Ive been trying to rm -rf individual folders, just randomly picking. It seems like they delete (the folders disappeared and you can hear disk io) and they just come back next boot.
Any other ideas? I’m about to smashy head the wall at this point 
I’ve been copying data out for two weeks between force reboots. I had been living with it up until it needed to replace a failed disk.
I just don’t want to wipe the whole thing. I do have the vast majority backed up but it’s still just a pain to do, especially given the size of most of the media files. Argh. All of my storage apps/media is all ZFS so everything flakes when the kernel starts panicking. I’m so sick of recovering jellyfins sqllite db 