The DQM process is killed / crashed on Windows while submitting many jobs from a remote node using a Logical Queue.
Most likely system running out of memory / pagefile due to the high load on the server
Release : 6.10
Component : DOLLAR UNIVERSE
In order to workaround the issue, restart manually DQM once it goes offline.
In order to fix the issue I would reduce the number of parallel jobs that can be launched simultaneously on the same server.
Increasing the memory and virtual memory of the Windows server would probably help, please check with your system admins why the system makes DQM process crash.