darrenfreeman

Results 42 comments of darrenfreeman

Kernel panic on incremental send/recv between two encrypted datasets on the same pool, dest is using zstd-19

Well, it happened again, under similar conditions: ``` zfs send rpool/src@snapshot | pv | zfs recv -u -o recordsize=1M -o compression=zstd-19 rpool/dest ``` The output from pv indicates that 21.3...

Kernel panic on incremental send/recv between two encrypted datasets on the same pool, dest is using zstd-19

Upon reboot, the pool status shows the same as above. A scrub has not begun, and the error shows the same, with `rpool/another/seventh/%recv`. The corresponding dataset had a `receive_resume_token`. I...

Kernel panic on incremental send/recv between two encrypted datasets on the same pool, dest is using zstd-19

Once again, after the second scrub, the error is no longer shown. I haven't rebooted, and I have been sending other datasets to this pool in the meantime.

Kernel panic on incremental send/recv between two encrypted datasets on the same pool, dest is using zstd-19

@Ornias1993 that was never conveyed to me. It's shipping in Debian without any warnings. Can you please link me to where it's been posted?

Kernel panic on incremental send/recv between two encrypted datasets on the same pool, dest is using zstd-19

FWIW, I've diffed at least a TB of data between snapshots mounted side by side, lz4 and zstd-19. So far no differences found, no data loss that I'm aware of.

Kernel panic on incremental send/recv between two encrypted datasets on the same pool, dest is using zstd-19

I have only seen public discussion in favour of zstd on zfs. If there are serious issues, then it needs to be given equal exposure. But for the record, is...

Kernel panic on incremental send/recv between two encrypted datasets on the same pool, dest is using zstd-19

For myself, and anyone trying to follow this, is it safe to leave datasets as zstd? Is the issue purely that sometimes non-raw datasets can't be received without rendering the...

Kernel panic on incremental send/recv between two encrypted datasets on the same pool, dest is using zstd-19

I am investigating without looking at any code. My current test is to recompress 3 TB, after dropping the L2ARC, SLOG, and stopping sanoid/syncoid. Many would find that a reasonable...

Kernel panic on incremental send/recv between two encrypted datasets on the same pool, dest is using zstd-19

Well it's currently sending that 3 TB dataset from lz4 to zstd-19, with practically nothing else happening on the box. Here is the resource usage reported by `top`: ``` load...

Kernel panic on incremental send/recv between two encrypted datasets on the same pool, dest is using zstd-19

Also @BrainSlayer it's doing 100-200 MB per *second*. I monitor this using `pv` in the pipe between send and receive.