You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@geronimo.apache.org by Beniamin Mazan <it...@mazan.pl> on 2008/06/13 14:39:47 UTC
Performance problem with Geronimo (2.1) / Entity Manager
Hi
My application listens for requests incoming via WebServices and then
processes it by sending something to JMS and database (postgres) in XA -
transaction.
I've noticed that after processing some thousands transactions - the time
required to finish transaction is being risen. I've looked to log and found
that in this case time is consumed by EM for execution of method
.remove(entity). After all geronimo stops responding for web requests and
need `kill -9` to shut down.
What does go wrong ?
thanks in advance
-----
thanks
Beniamin
--
View this message in context: http://www.nabble.com/Performance-problem-with-Geronimo-%282.1%29---Entity-Manager-tp17822465s134p17822465.html
Sent from the Apache Geronimo - Users mailing list archive at Nabble.com.
Re: Performance problem with Geronimo (2.1) / Entity Manager
Posted by Beniamin Mazan <it...@mazan.pl>.
Kevan Miller wrote:
>
>
> On Jun 13, 2008, at 8:39 AM, Beniamin Mazan wrote:
>
>>
>> Hi
>> My application listens for requests incoming via WebServices and then
>> processes it by sending something to JMS and database (postgres) in
>> XA -
>> transaction.
>> I've noticed that after processing some thousands transactions - the
>> time
>> required to finish transaction is being risen. I've looked to log
>> and found
>> that in this case time is consumed by EM for execution of method
>> .remove(entity). After all geronimo stops responding for web
>> requests and
>> need `kill -9` to shut down.
>>
>> What does go wrong ?
>
> Hi Beniamin,
> Sounds like we've got a memory leak...
>
> Is this G 2.1.1?
> Do you have a sample app that can demonstrate the problem?
> Do you have the stack trace for the remove(entity) call, that you're
> observing?
> Running with the following JAVA_OPTS should yield additional info...
>
> JAVA_OPTS="-Xmx512m -XX:MaxPermSize=128m -verbose:gc -XX:
> +PrintGCDetails -XX:+HeapDumpOnOutOfMemoryError"
>
> Set the heap and permgen sizes as appropriate to your application
> environment. The smaller the better... The PrintGCDetails will
> indicate if we losing memory. If you are running out of memory, a heap
> dump which we can use to help diagnose.
>
> --kevan
>
>
Hi, thanks for your reply, Kevan. I'll try to follow your advice in the very
first days of next week and i'll notify you if problem is with memory leak.
On the other hand, I didn't get any OutOfMemory exception till now.
-----
thanks
Beniamin
--
View this message in context: http://www.nabble.com/Performance-problem-with-Geronimo-%282.1%29---Entity-Manager-tp17822465s134p17831177.html
Sent from the Apache Geronimo - Users mailing list archive at Nabble.com.
Re: Performance problem with Geronimo (2.1) / Entity Manager
Posted by Kevan Miller <ke...@gmail.com>.
On Jun 13, 2008, at 8:39 AM, Beniamin Mazan wrote:
>
> Hi
> My application listens for requests incoming via WebServices and then
> processes it by sending something to JMS and database (postgres) in
> XA -
> transaction.
> I've noticed that after processing some thousands transactions - the
> time
> required to finish transaction is being risen. I've looked to log
> and found
> that in this case time is consumed by EM for execution of method
> .remove(entity). After all geronimo stops responding for web
> requests and
> need `kill -9` to shut down.
>
> What does go wrong ?
Hi Beniamin,
Sounds like we've got a memory leak...
Is this G 2.1.1?
Do you have a sample app that can demonstrate the problem?
Do you have the stack trace for the remove(entity) call, that you're
observing?
Running with the following JAVA_OPTS should yield additional info...
JAVA_OPTS="-Xmx512m -XX:MaxPermSize=128m -verbose:gc -XX:
+PrintGCDetails -XX:+HeapDumpOnOutOfMemoryError"
Set the heap and permgen sizes as appropriate to your application
environment. The smaller the better... The PrintGCDetails will
indicate if we losing memory. If you are running out of memory, a heap
dump which we can use to help diagnose.
--kevan