possible releng13 deadlock ?

From: mike tancsa <mike_at_sentex.net>
Date: Tue, 22 Jun 2021 13:49:06 -0400
I was trying out a zfs send | recv on a new controller this morning
using the latest RELENG_13 upto the last commit there    
(25302a721f69d34c72d8bd3e0c316ad12647e281) and experienced a hard lock
on the system.  The motherboard is just a desktop board, so no serial
hooked up nor ipmi.  However, reverting to a kernel from the 20th
(stable/13-659f77766) seems to be OK.  Note, this was one try deadlock
so cant be sure if it was a fluke and the issue is from some time else.

But I will try and repeat a few sends (approx 20min) with the working
kernel and then try again with the latest kernel.  Anyone else using the
latest 13 stable and ZFS from today notice anything ?

I am just sending a zfs dataset from one set of striped disks to
another.  On the plus side, its FAST!

 zpool iostat testmdt 1
              capacity     operations     bandwidth
pool        alloc   free   read  write   read  write
----------  -----  -----  -----  -----  -----  -----
testmdt     1.31T  1.41T      1  4.78K  6.62K   602M
testmdt     1.31T  1.40T      0  6.51K      0   819M
testmdt     1.31T  1.40T      0  6.52K      0   820M
testmdt     1.32T  1.40T      0  6.52K      0   820M
testmdt     1.32T  1.40T      0  6.51K      0   819M
testmdt     1.32T  1.40T      0  8.22K      0  1.02G
testmdt     1.32T  1.40T      0  4.98K      0   622M
testmdt     1.32T  1.40T      0  8.58K      0  1.06G
testmdt     1.32T  1.40T      0  9.18K      0  1.13G
testmdt     1.32T  1.40T      0  1.44K      0   172M
testmdt     1.32T  1.40T      0  1.19K      0   153M
testmdt     1.32T  1.40T      0  5.21K      0   653M
testmdt     1.32T  1.40T      0  4.09K      0   524M
testmdt     1.32T  1.40T      0  8.72K      0  1.06G
testmdt     1.32T  1.40T      0  2.51K      0   321M
testmdt     1.32T  1.39T      0  10.5K      0  1.29G
testmdt     1.32T  1.39T      0  6.42K      0   808M



Received on Tue Jun 22 2021 - 17:49:06 UTC

Original text of this message