Hey Roger, I tried the same (-XX:MaxPermSize=256M) and guess what, no more crashes, no more deadlocks. I would love to have had error messages which at least point in the right direction instead of "bean not found", "response already commited" and "broken pipe"... Now I know that an OutOfMemory exception in tomcat doesn't mean that you are out of memory, and a response already commited/client aborted/broken pipe -> does.
Thanx for the hint Roger. regards Leon On 3/18/06, Roger Keays <[EMAIL PROTECTED]> wrote: > > > I've studied a number of stack traces and the locking seems fine - there > > is evidence of cyclical deadlock. The problem is that a whole lot of > > threads (including the ones holding the locks) are listed as 'waiting > > for monitor' but I can't see why. > > I found out that it was the garbage collector that those threads were > waiting on, but the gc had given up because I was out of PermGen memory > space. > > -XX:MaxPermSize=128m fixes the issue :) > > -- > ---------------------------------------- > Ninth Avenue Software > p: +61 7 3870 8494 (UTC +10) > f: +61 7 3870 8491 > w: http://www.ninthavenue.com.au > e: [EMAIL PROTECTED] > ---------------------------------------- > > > --------------------------------------------------------------------- > To unsubscribe, e-mail: [EMAIL PROTECTED] > For additional commands, e-mail: [EMAIL PROTECTED] > > --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
