Thank you very much for answering.
> The in doubt /failure suspected signals are generated when instances
> (including DAS) do not send out heartbeats (...)
> Could you double check that the DAS process and instance process have
> actually failed using jps on the machine where both DAS and instance reside?
Well the damn thing insists on happening overnight, so I'm always involved too late, at autopsy time...
But the processes really [i]have[/i] vanished (I noted the PIDs at launch time and they're no more there the day after)...
> Also check if you have sufficient heap set up to
> handle the load (albeit moderate load) and have sufficient available system
> RAM.
It seems so. I monitor the system and process RSS, and the system swap. I also have GC logs enabled, and they show the instance mem usage doesn't even come close to its -Xmx setting.
> It would help to know what specific build you are on
> along with other details such as JDK version, system information, OS platform, whether on virtual machines, etc.
You're right.
I was naively expecting that this was a known issue that escaped my searches, but since it appears not to be the case, I have to provide more info...
GF build: 2.1 b60E
JRE version: Sun's JDK, 1.6.0_13-b03
System information: RHEL redhat-release-4AS-8.0.el4_7.1, running in a VMWare Server
[Message sent by forum member 'jduprez' (jduprez)]
http://forums.java.net/jive/thread.jspa?messageID=347686