ZFS issues on 13-current snapshot

[Matt Churchyard]

Hello

Just a follow up on my email yesterday about ZFS issues on a 13-current snapshot.

I have now scrapped the 13 system, reinstalled 12.2 on a new pool, and copied the same ~400GB of data to it.
This is the outcome -

1) I no longer see phantom data in snapshots that have been sent to the pool. Every snapshot lists 0 usage. "zfs diff" is also much happier and runs in seconds on any dataset.

2) I can scrub the entire pool in 30 minutes with the disks pushing 150MBps+ pretty much the entire time. I eventually cancelled the scrub on 13; It stayed at 19% for about 4 hours, then made some progress to about 65%, before sitting there for another 6 hours. I gave up in the end and stopped it.

# zpool list
NAME      SIZE  ALLOC   FREE  CKPOINT  EXPANDSZ   FRAG    CAP  DEDUP  HEALTH  ALTROOT
offsite  14.5T   906G  13.6T        -         -     0%     6%  1.00x  ONLINE  -
# zpool status
  pool: offsite
 state: ONLINE
  scan: scrub repaired 0 in 0 days 00:27:46 with 0 errors on Thu Jan 28 13:44:47 2021

3) Additional minor thing unrelated to ZFS or storage, but I noticed that 'systat -ifstat 1' was showing 23MB/s throughput which let me to believe I must have been running at a gigabit and just getting slow throughput for other reasons. I probably should have looked into this more considering the sender showed 11MB/s. On 12.2, I see 11MB/s on both ends, making it very clear it was definitely capping at 100Mbps. I eventually tracked down a 100Mbps link - obviously not a FreeBSD issue, but it seems systat was reporting double on 13.

Maybe some of these are due to it being a current dev snapshot and having various debug features active, but I think I will be seeing the general response when 13 comes out before I go anywhere near it...

Regards,
Matt Churchyard