maxBufferedDocs vs autoCommit->maxDocs

["Bouis, Laurent" <La...@sonyconnect.com>]

 

Hi,

 

What is the difference between the
<maxBufferedDocs>1000</maxBufferedDocs>, and the
<autoCommit><maxDocs>1000</maxDocs>...</autoCommit> parameters in
solrconfig.xml? 

Do they influence the frequency of flush to disk and document
distribution in segments in a different way?

 

When I did some test with low identical values, I saw similar behavior
in terms of frequency of index data flushed to disk (although the number
of segments sometimes looked a bit different).

 

 

Thanks.

 

 

Laurent

Re: maxBufferedDocs vs autoCommit->maxDocs

[Mike Klaas <mi...@gmail.com>]

On 27-Sep-07, at 3:35 PM, Bouis, Laurent wrote:

> What is the difference between the
> <maxBufferedDocs>1000</maxBufferedDocs>, and the
> <autoCommit><maxDocs>1000</maxDocs>...</autoCommit> parameters in
> solrconfig.xml?
>
> Do they influence the frequency of flush to disk and document
> distribution in segments in a different way?

maxBufferedDocs affects disk flushing behaviour, but is purely a  
lucene-level thing that might affect performance and memory usage,  
but not semantics.

autoCommit is a Solr-level option that affects when added document  
become visible to searches (very important to search semantics).  A  
side-effect of a commit is a lucene-level flush, so if  
maxBufferedDocs > autoCommit it does not have any effect.

cheers,
-Mike