up vote 0 down vote favorite
share [g+] share [fb]

We have a RedHat server whose only job is to run a JBoss server. Monitors said that memory usage spiked (we have the JVM limited to far less than the total memory on the system) and JBoss crashed. We restarted and everything seems ok now.

Odd thing is that sar data for 10 min leading up to the crash is simply not there. Load average got up to the 50s. I have seen severely busy systems (350+ Load Average) still collect sar data. Does anyone have any idea what could cause sar to stop collecting data?

link|improve this question
67% accept rate
feedback

1 Answer

up vote 0 down vote accepted

Turns out we were out of memory and swap space. Question now becomes (for another thread) what would cause a JBoss instance take up at least 4 more GB of memory than it is allocated in perm space and heap?

link|improve this answer
I suggest you open a new question. – Insyte Apr 9 '10 at 21:01
I did. Thus the 'for another thread' parenthetical. :) – Ichorus Apr 9 '10 at 21:32
feedback

Your Answer

 
or
required, but never shown
discard

Not the answer you're looking for? Browse other questions tagged or ask your own question.