Hi Jukka,
Thank you for your immediate response.
The OOM error does not always occur, but once the OOM error occurs the
system is getting slower. And I didnt set the maxLength, so the
default value would be 10000.
As you did guess, there were huge documents indexed some are several
hundreds mega bytes per file. I think this causes the error.
In this case, how can i get the indices normal? Any help would be
really helpful.
Thanks.
BR,
Marcus
Thanks!
On Feb 17, 2010, at 4:44 PM, Jukka Zitting <[email protected]>
wrote:
Hi,
On Wed, Feb 17, 2010 at 3:53 AM, MINT <[email protected]> wrote:
Nowadays, I get java.lang.OutOfMemoryError every 1 or 2 days like
below..
[...]
I'm running Jackrabbit 2.0 on Tomcat5.5 with max heap memory 1024m..
I don't have any clue why this happens occasionally..
Is there an any way to fix indices or any option not to this happen?
Does the OOM error always occur with the index merger? What's your
maxFieldLength setting? It may be that you have some huge document(s)
indexed, which together with a large maxFieldLength setting may cause
extra memory use during index merging.
Alternatively you may simply be hitting a normal memory limit
especially if you have lots (hundreds) of open sessions accessing the
repository. Increasing the max heap limit should get you around that.
You may also want to try getting a heap dump of the JVM when you see
the memory use go over some limit (80% of max heap). That should help
identify where the memory is going.
BR,
Jukka Zitting
