java out of memory errors crashing system

 2 Replies
 0 Subscribed to this topic
 27 Subscribed to this forum
Sort:
Author
Messages
elessarfl
Basic Member
Posts: 4
Basic Member
    We are experiencing sporadic (a few times a month) issue where multiple javacore dumps, heapdumps and Snap traces are written in rapid succession (a few minutes) to the $LAWDIR/system directory. This is quickly followed by the system freezing and the only way to recover is to bounce the Env. Is anyone else experiencing this? We are LSF 9.0.0.5 patched up to current, Apps 9.0.0.4 on UNIX with AIX 5.3 and Oracle 10g and WebSphere 6.0.2.35 and MS ADAM.
    I have had several Lawson Cases open on this and they have not been able to resolve. Thanks
    Xin Li
    Veteran Member
    Posts: 133
    Veteran Member
      Have you set Initial Heap Size and Maximum Heap Size and enable Verbose Garbage Collection? You can find these setting under "application servers-->you servername-->Process definition --Javal Virtual Machine".
      EricS
      Veteran Member
      Posts: 80
      Veteran Member
        We've had the same problems with a large data set sorting in JVM. It used to just crash the system, we vertically scaled with 3 app servers. It made it better in that just one JVM woud fill up and crash one app, but either query wizard or a user filtering a large data set by a non-indexed field in a drill around dialog box will still crash the application. Of course once a user blows up once, they abort query wizard and just start another one.
        ESP 6 is supposed to make this more palatable by bringing data back in pages instead of trying to sort a large data set in Java memory. We did have something lock the whole system up today, but by the time it lets me in to look at what's going on, its starting to recover with out posting any error messages. If anyone can figure out a way to get error messages to post, I'd be very interested in that.