You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@jackrabbit.apache.org by Marcel Reutegger <ma...@gmx.net> on 2009/08/03 09:13:08 UTC

Re: Disable Indexing but still getting IndexMerger log messages

Hi,

2009/7/29 Paulo Silva <al...@gmail.com>:
> I have a big "copy" operation in my application that I'm trying to optimize.
> For that I disabled indexing by changing my index.xml to have
> "<index-rule nodeType="nt:base"></index-rule>".
>
> Still, I'm getting log messages like this in my copy operation:
> 12:30:23,921 INFO  [IndexMerger] merged 20 documents in 15 ms into _5ql.

this won't disable indexing entirely. it will just limit the number of
properties indexed to a minimum. that is, jcr:primaryType,
jcr:mixinTypes and any kind of reference properties are always
indexed. it is not possible to exclude nodes of a certain type or a
sub tree from indexing. hierarchical information (parent-child
relations) are also resolved within the index and requires that all
nodes are present in the index.

regards
 marcel

Re: Disable Indexing but still getting IndexMerger log messages

Posted by Marcel Reutegger <ma...@gmx.net>.
On Tue, Aug 4, 2009 at 14:49, Paulo Silva<al...@gmail.com> wrote:
> Thanks for the explanation.
>
> As I stated, I have a big "copy" operation (1000-5000 nodes) in my
> application. Is there any way to optimize this?
> I've playing with configs and increasing minMergeDocs to 2000 seems to
> help (is there any problem increasing this?),

it will use more memory, but otherwise isn't really a problem.

> but still takes around
> 15s in my machine.

hmm, is that just the time to index the copied content or the overall
copy operation?

> Any tips to improve this copy?

you could disable the index altogether, but I guess that's not what you want...

do you have nt:resource nodes in that content, with expensive text
extractors configured? pdf text extraction with pdfbox is known to be
expensive.

regards
 marcel

Re: Disable Indexing but still getting IndexMerger log messages

Posted by Paulo Silva <al...@gmail.com>.
Thanks for the explanation.

As I stated, I have a big "copy" operation (1000-5000 nodes) in my
application. Is there any way to optimize this?
I've playing with configs and increasing minMergeDocs to 2000 seems to
help (is there any problem increasing this?), but still takes around
15s in my machine.

Any tips to improve this copy?


Thanks in advance



On Mon, Aug 3, 2009 at 8:13 AM, Marcel
Reutegger<ma...@gmx.net> wrote:
> Hi,
>
> 2009/7/29 Paulo Silva <al...@gmail.com>:
>> I have a big "copy" operation in my application that I'm trying to optimize.
>> For that I disabled indexing by changing my index.xml to have
>> "<index-rule nodeType="nt:base"></index-rule>".
>>
>> Still, I'm getting log messages like this in my copy operation:
>> 12:30:23,921 INFO  [IndexMerger] merged 20 documents in 15 ms into _5ql.
>
> this won't disable indexing entirely. it will just limit the number of
> properties indexed to a minimum. that is, jcr:primaryType,
> jcr:mixinTypes and any kind of reference properties are always
> indexed. it is not possible to exclude nodes of a certain type or a
> sub tree from indexing. hierarchical information (parent-child
> relations) are also resolved within the index and requires that all
> nodes are present in the index.
>
> regards
>  marcel
>