CORBA Memory Leak

Bill Murray bmurray at snf.stanford.edu
Tue Mar 13 08:37:09 PST 2001


John and Mike,

>From the logs, it appears that the event server ran out of memory at
about 7:15am this morning.  All the other servers appeared to be running
great; events just weren't being posted any longer.   The servers had
been up for one week.   However, it was a week of very heavy activity
where we frequently exceeded 50 concurrent active users.  As soon as I
finish the database/code cleanup, I will double the memory allocated to
the event server JVM which should give us a couple of weeks under a
heavy load.

Of course, I immediately restarted all the servers when I found the
problem this morning.

Status on database cleanup:  I'm currently modifying the ticket generation
code to handle our new default project/account changes.  After this, I
need to make a quick set of global changes to accommodate the change
from expired to active in works_on and charges_to.  Then (finally), we're
ready to modify all the tables and test extensively.

Bill 




More information about the coral mailing list