On Dec 1, 2006, at 8:58 AM, Alexandru Popescu wrote:
On 11/29/06, Tom Duffey <[email protected]> wrote:
On Nov 28, 2006, at 8:19 PM, Alexandru Popescu wrote:
> On 11/29/06, Tom Duffey <[email protected]> wrote:
>> Hi All,
>>
>> Has anyone ran any tests to see how or if memory usage changes
with
>> the different Jackrabbit storage options? I've heard that bdb
uses
>> less RAM than Derby, but what about using a relation database
>> backend? Does that place less memory demand on the application
>> servers?
>>
>> We have a 2.1.3 site using BDB w/768MB heap and it is constantly
>> running out of memory and crashing so I'm looking for options
that do
>> not require more hardware. Any ideas?
>>
>
> Have you identified why are you getting the OOM? It looks pretty
weird
> and I am quite sure this is not the fault of BDB.
No, it just happens every few days of normal site usage. It's a
semi-
busy site that pushes about 75GB data per month. It does not seem to
be triggered by activation or any other Magnolia admin action.
Tom
Are you performing updates of the deployed application? Or deploying
more JSPs files while running? As I said, I find it quite weird,
because I haven't seen it before. Is it a PermGen memory error or a
normal OutOfMemory error?
No, we do not deploy anything without restarting Tomcat. However, in
our latest crash I noticed that the OOM is first caused by Lucene:
ERROR org.apache.jackrabbit.core.query.lucene.IndexMerger
IndexMerger.java(run:301) 27.11.2006 15:23:05 Error while merging
indexes: java.lang.OutOfMemoryError
Unfortunately, I lost the previous crash log so can't check if this
is how it always starts. We don't use Lucene for anything, could I
export the site, turn of indexing, and reload?
Tom
----------------------------------------------------------------
for list details see
http://www.magnolia.info/en/magnolia/developer.html
----------------------------------------------------------------
