A background method server doesn't have foreground connections.
A method context is the method server's term for a body of work. For
foreground requests, there's a method context for each RMI request, for
instance. The same concept applies to background work, e.g. queue tasks.
Trying to do 74 things at once but not managing to use much CPU is a bad
sign. I'm inclined to suspect a deadlock. I assume the server status
page isn't noting a deadlock, but that just means there's no simple
thread deadlock. This could well be a more complex deadlock -- for
instance, one where 'n' threads holding database connections are all
waiting for a lock that's held by another thread which is itself waiting
for a database connection.
The key here is to get and analyze thread dumps for the method server in
question. Unfortunately at R9.1 we don't yet add extra information
about database connection usage, etc, to JMX-based thread dump output --
as we do in more recent releases. Thus even with a thread dump, it may
be a bit tricky to find the culprit.
--
Jess Holle
