bmx health daemon
It can happen that bmx6 crashes due to bugs and unforseen scenarios. If such events happen seldom they are likely to be overseen unless (according to Murphy's) its showtime in a productive deployments.
One option to improve the chances of finding such bugs is
a daemon in charge of monitoring the health of bmx6 process.
This daemon should be started/stopped by qmp's /etc/init.d/bmx6 start/stop functions.
Whenever it detects that bmx6 process has died it should save (append) the system log file (logread) to a permanent file (eg. /root/bmx6-errors.log) without overwriting previous errors.
Then it should restart the daemon.
Optionally, send a nice diplomatic email to the bmx6 author or node owner saying that something f*** up!
The daemon must be stopped when bmx6 is stopped on purpose.