You are viewing a plain text version of this content. The canonical link for it is here.
Posted to java-user@lucene.apache.org by "Aigner, Thomas" <TA...@WescoDist.com> on 2006/10/26 17:18:33 UTC
Possible memory issue?
Howdy all,
I have a issue with java running out of memory after the search
has been running for a while. We are using 1.9.1 release and I check
the indexreader's version to determine if I need to get a new searcher
for searches (so I pick up any changes to the index). I am seeing jumps
in memory for the first search after loads to the index. After these
jumps the memory does not come back down after a load has taken place..
more loads = more memory consumed until we eventually run out. Is it
possible that this is a bug with lucene caching the memory from the last
search and creating memory for the next search? Is there a way to tell
lucene to drop it's cached memory if that is the case?
Has anyone come across this possible problem with multiple sort
fields?
Ex. Java memory before loads hovers around 110M and about half of it
free.
Many searches take place, memory stays relatively the same
Then Load takes place
Next search uses 192M memory with 80Meg free..
---------------------------------------------------------------------
To unsubscribe, e-mail: java-user-unsubscribe@lucene.apache.org
For additional commands, e-mail: java-user-help@lucene.apache.org
RE: Possible memory issue?
Posted by "Aigner, Thomas" <TA...@WescoDist.com>.
Thanks for the advice drj. I do close the searcher and set it to null
before instantiating another searcher. I believe that I am closing the
reader and writer at the correct times as well...
-----Original Message-----
From: d rj [mailto:drjat320@gmail.com]
Sent: Thursday, October 26, 2006 11:40 AM
To: java-user@lucene.apache.org
Subject: Re: Possible memory issue?
I would suggest checking that close() is properly called for all
IndexSearcher/Reader/Writer objects when doing adds/deletes and when
recreating IndexSearcher object.
Free memory in the JVM heap can diminish quickly if these objects aren't
properly disposed.
-drj
On 10/26/06, Aigner, Thomas < TAigner@wescodist.com> wrote:
>
> Howdy all,
>
> I have a issue with java running out of memory after the
search
> has been running for a while. We are using 1.9.1 release and I check
> the indexreader's version to determine if I need to get a new searcher
> for searches (so I pick up any changes to the index). I am seeing
jumps
> in memory for the first search after loads to the index. After these
> jumps the memory does not come back down after a load has taken
place..
> more loads = more memory consumed until we eventually run out. Is it
> possible that this is a bug with lucene caching the memory from the
last
> search and creating memory for the next search? Is there a way to
tell
> lucene to drop it's cached memory if that is the case?
>
> Has anyone come across this possible problem with multiple
sort
> fields?
>
>
> Ex. Java memory before loads hovers around 110M and about half of it
> free.
> Many searches take place, memory stays relatively the same
> Then Load takes place
> Next search uses 192M memory with 80Meg free..
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: java-user-unsubscribe@lucene.apache.org
> For additional commands, e-mail: java-user-help@lucene.apache.org
>
>
---------------------------------------------------------------------
To unsubscribe, e-mail: java-user-unsubscribe@lucene.apache.org
For additional commands, e-mail: java-user-help@lucene.apache.org
Re: Possible memory issue?
Posted by d rj <dr...@gmail.com>.
I would suggest checking that close() is properly called for all
IndexSearcher/Reader/Writer objects when doing adds/deletes and when
recreating IndexSearcher object.
Free memory in the JVM heap can diminish quickly if these objects aren't
properly disposed.
-drj
On 10/26/06, Aigner, Thomas < TAigner@wescodist.com> wrote:
>
> Howdy all,
>
> I have a issue with java running out of memory after the search
> has been running for a while. We are using 1.9.1 release and I check
> the indexreader's version to determine if I need to get a new searcher
> for searches (so I pick up any changes to the index). I am seeing jumps
> in memory for the first search after loads to the index. After these
> jumps the memory does not come back down after a load has taken place..
> more loads = more memory consumed until we eventually run out. Is it
> possible that this is a bug with lucene caching the memory from the last
> search and creating memory for the next search? Is there a way to tell
> lucene to drop it's cached memory if that is the case?
>
> Has anyone come across this possible problem with multiple sort
> fields?
>
>
> Ex. Java memory before loads hovers around 110M and about half of it
> free.
> Many searches take place, memory stays relatively the same
> Then Load takes place
> Next search uses 192M memory with 80Meg free..
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: java-user-unsubscribe@lucene.apache.org
> For additional commands, e-mail: java-user-help@lucene.apache.org
>
>