lucene-solr-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Jack Krupansky" <j...@basetechnology.com>
Subject Re: Reindex Issues
Date Tue, 25 Nov 2014 12:03:07 GMT
When a document is reindexed, the old document is deleted and the new 
document is added. The deleted document is not visible on queries, but the 
document frequency (df) for terms includes the count of deleted documents 
containing the terms. I would expect that df would double if all documents 
are reindexed and no segment merges occur. Once segment merges occur, the 
deleted documents are "expunged" and df will once again be more accurate - 
at least in terms of the segments that were merged. An optimize essentially 
does a merge on all of the segments, either combining segments or rewriting 
them.

The constant score should normally be 1.0. If it is not, maybe you have 
query boost terms, and they are using the df of the boost terms.

-- Jack Krupansky

-----Original Message----- 
From: Ahmet Arslan
Sent: Tuesday, November 25, 2014 4:50 AM
To: solr-user@lucene.apache.org
Subject: Re: Fwd: Reindex Issues

Hi,

The query you use is constant score query, so as long as all documents 
assigned to same score, it is not a problem.


Also you may want to read about expungeDeletes.

Ahmet


On Tuesday, November 25, 2014 10:23 AM, rashi gandhi 
<gandhirashi19@gmail.com> wrote:
Hi,



I have created two shards at solr server and around 4K documents are
equally indexed over these two shards.

I did re-indexing for all the indexed documents (updating exiting docs with
same data again).

After Re-indexing, I found that my indexes are not optimized and there is
change in the score of documents.



Previously when I Query *:* = returned all docs with same score (let’s say
= 0.046)

After Re-indexing, when I query *:*  = returned all docs but with different
and sudden high score (Let’s Say for first 3K docs = 0.46 and for last 1K =
0.45)



I am not able to understand why there is sudden increase as well as change
in the score of docs after re-indexing.

Please provide me some pointers on this.



Also, I wanted to know if I re-index small number of documents, why I am
not able to see counts for deleted docs from UI and indexes are also
optimized ??



Thanks,

Rashi 


Mime
View raw message