|
Home > Archive > WebSphere Application Server > March 2004 > OutOfMemory Errors
You are viewing an archived Text-only version of the thread.
To view this thread in it's original format and/or if you want to reply to
this thread please [click here]
| Author |
OutOfMemory Errors
|
|
|
| Regarding getting these things tracked down, what do you guys do
that's outside the MustGather advice that IBM posts off their support
web? Right now, I've got a PMR open with 'em that has me basically
waiting for an OOM condition to return - after having bumped an
application's maxheap to 768m (unheard of, as the same app has run
just fine with 256m for about 18 months), and what I'm looking to
avoid - is getting a song & dance from IBM about how they don't have a
Solaris-based heaptool that's ready for primetime.
We've got our application folks looking into the application with
jprobe, but if anyone's got some other things we can try...additional
data we can gather...I'm all ears. WebSphere 4.06, running on Solaris
8.
Thanks in advance! -- Tony
| |
|
| Are you able to reproduce the problem and have you a test case to cause the
problem ? Are you sure there is no leakage, like an ever-growing applicative
cache of data ? Did you load and stress test the system ? Have you looked at
the output of the "verbosegc" command line argument
(http://www-106.ibm.com/developerwor...rary/i-garbage3) ? Did you
watch the evolution of the memory usage with the Resource Analyzer ? You may
also want to activate -XrunpmiJvmpiProfiler command line argument
(http://publib.boulder.ibm.com/infoc...opic=/com.ibm.w
ebsphere.nd.doc/info/ae/ae/rprf_datacounter4.html) and watch again with RA.
There is (too ?) much to do in such a situation. Unless you have a test case
and can cause the problem, I don't think a profiler is the right tool to
begin with (but it's only based on my own experience).
| |
| Andreas Nagel 2004-03-31, 1:34 pm |
| We got this kind of problem when we increased the PreparedStatementCache in
our datasources.
See
http://www-1.ibm.com/support/docvie...tf-8&lang=en+en
Hope this helps,
Andreas
"TF" <tfarne@yahoo.com> schrieb im Newsbeitrag
news:1u7u50h5nbveirmi2a03f9n4ihkpkb47n9@
4ax.com...
Regarding getting these things tracked down, what do you guys do
that's outside the MustGather advice that IBM posts off their support
web? Right now, I've got a PMR open with 'em that has me basically
waiting for an OOM condition to return - after having bumped an
application's maxheap to 768m (unheard of, as the same app has run
just fine with 256m for about 18 months), and what I'm looking to
avoid - is getting a song & dance from IBM about how they don't have a
Solaris-based heaptool that's ready for primetime.
We've got our application folks looking into the application with
jprobe, but if anyone's got some other things we can try...additional
data we can gather...I'm all ears. WebSphere 4.06, running on Solaris
8.
Thanks in advance! -- Tony
|
|
|
|
|