[debian-mysql] Bug#991399: mariadb: Daemon hangs on stop with replication and parallel slaves

[Guillem Jover]

Source: mariadb-10.5
Source-Version: 1:10.5.10-2
Severity: serious
Forwarded: https://jira.mariadb.org/browse/MDEV-20821
Control: fixed -1 1:10.5.11-1~exp1

Hi!

When replication is enabled with the slave running in parallel mode,
the daemon completely hangs and will not terminate, which means daemon
shutdowns need to timeout and a SIGKILL sent to take effect, as such
daemon restarts will fail.

This is fixed in upstream 10.5.11, the relevant upstream bugs are:

  https://jira.mariadb.org/browse/MDEV-20821
  https://jira.mariadb.org/browse/MDEV-22370

I've set this as serious, as ending up with no mariadb running on a
system can be pretty fatal, and because upstream had it as "critical",
but feel free to lower if you think it would not deserve it. Given that
there appears to be a version that fixes this issue in experimental, it
seems this would not be too cumbersome, and it would be nice if that
could be targeted for bullseye.

Thanks,
Guillem