MaxScale crashed without much of a trace. It is used in a dev environment with some web sites, but its overall load is fairly low. Configuration is trivial, 1 master + 1 slave, RWsplit and nothing else.

Version:
[root@cgdsqlmax1 maxscale]# rpm -q maxscale
maxscale-2.5.4-1.rhel.8.x86_64

maxscale.log (this is the last line in the log; the second to last was from hours before that):
double free or corruption (!prev)

systemd journal:
Oct 20 16:15:29 cgdsqlmax1 systemd[1]: maxscale.service: Watchdog timeout (limit 1min)!
Oct 20 16:15:29 cgdsqlmax1 systemd[1]: maxscale.service: Killing process 930 (maxscale) with signal SIGABRT.
Oct 20 16:16:59 cgdsqlmax1 systemd[1]: maxscale.service: State 'stop-sigabrt' timed out. Terminating.
Oct 20 16:18:29 cgdsqlmax1 systemd[1]: maxscale.service: State 'stop-sigterm' timed out. Killing.
Oct 20 16:18:29 cgdsqlmax1 systemd[1]: maxscale.service: Killing process 930 (maxscale) with signal SIGKILL.
Oct 20 16:18:29 cgdsqlmax1 systemd[1]: maxscale.service: Main process exited, code=killed, status=9/KILL
Oct 20 16:18:29 cgdsqlmax1 systemd[1]: maxscale.service: Failed with result 'watchdog'.

Other notes: MaxScale 2.4 was fairly stable at the same place for a year or so. A crash on a lightly loaded environment just 2 weeks after the upgrade puts a production upgrade under question.

Would be nice to get at least Restart=always in systemd unit file.

Attaching the config and the monitoring graph with connections prior to the crash.