Repeatable database corruption with mariadb1011-server 10.11.8

[David Brownlee <abs%absd.org@localhost>]

I'm seeing non fixable database corruption on mariadb1011-server since
the update to 10.11.8

(non fixable in the sense that mariadb cannot fix it, and starting up
innodb_force_recovery=7 to force readonly, then dump/restore... or a
full wipe and reload is required)

This is on NetBSD-10/amd64

Prior to https://github.com/MariaDB/server/pull/3253 being added to
the pkgsrc patches I would see "/etc/rc.d/mariadb stop" hang every
time, leaving a corrupted database almost every time.

Since that patch was added to pkgsrc "/etc/rc.d/mariadb stop" has
stopped hanging, but I've had the test databases on my laptop
corrupted three times (rebuilding ~36GB of database being somewhat
time consuming). After the second time I switched back to 10.11.7
binaries, and ran without issue for a while, then back to the latest
10.11.8 binaries and had corruption on the next but one
shutdown/restart cycle.

I'm about to switch back to 10.11.7 built against the latest pkgsrc to
confirm the issue is still not present there.

Is anyone else using mariadb1011-server 10.11.8 on NetBSD, and have
they seen any issues?

Thanks

David