jackrabbit-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Tomasz Łazarecki <tomasz.lazare...@gmail.com>
Subject Re: Out Of Memory Error while indexing
Date Tue, 09 Feb 2010 10:38:56 GMT
That's only an example not the greatest solution.

So, here we go...
1.I would register my nodeType VersionControlNode.
2.Next i would add two content props: one for the smaller and second for the
bigger files in the VersionControlNode.
3.I'm providing the client api so i would check
if(fileSize>MAXIMUM_SMALL_FILE_SIZE) then i'm putting content into the
jcr:bigFIle, else to the
4. Because Jackrabbit can show to lucene which content should be indexed and
analayzed i'm doning it via the proper configuration. By that i mean
jcr:smallFile is indexed, jcr:bigFile is not.

As i get it from your e-mail, you are providing the client api.
ps:I don't find a solution for the defualt configuration...

Best regards/Pozdrawiam
Tomasz Łazarecki

2010/2/9 <mail@jutzig.de>

> Hi Tomasz,
> thank you for your reply. Please see comments below
> > One of the solution i have on my mind is:
> > 1.you may register your own type like: nt:BigAs**File of course the name
> > should be different.
> > 2. register it on server side via coping file custom_nodetypes.xml  to
> the
> > ${repo.home}/repository/nodetypes
> > 3. set to not index field of type jcr:bigAs**content
> >
> Like stated earlier, I'm using Jackrabbit as a version control system. The
> client is implemented as an eclipse team provider similar to the build-in
> CVS team provider. Now if a user commits a file test.txt with just a few
> bytes of content, I do want that to be indexed. So I create a nt:file, set
> the contents, and set the mime type. Later the user changes test.txt and
> commits it. This time the test.txt has 500MB. It's still the same node
> (nt:file) and I'd rather not delete the node and add a new one of another
> kind.
> Now the question for me is, how can I prevent Jackrabbit from crashing on
> these large files?
> I have seen the wiki links you gave me before, but so far I wasn't able to
> apply that to my use case.
> How would a configuration look like for
> 'stick to the default settings, but don't index properties larger than N
> bytes?'
> And if that doesn't work, how would I write:
> 'index everything except jcr:data where jcr:mimeType equals my:bigAssFile?'
> Thanks for the support and best regards,
> Johannes

  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message