lucene-java-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From msftbl...@aol.com
Subject Re: Java Indexer + DotLucene + IIS question
Date Mon, 31 Oct 2005 20:36:29 GMT
Never seen this product...but in my case I would have the indexer running on one webserver
and the second it creates a file, Replistor will take care of getting to all the other machines...but
will there be any issues with compressing etc.? I have come across where .cfs files were missing
etc. 
 
-----Original Message-----
From: Wesley MacDonald <wes@extremesoftware.ca>
To: java-user@lucene.apache.org
Sent: Mon, 31 Oct 2005 15:09:34 -0500
Subject: RE: Java Indexer + DotLucene + IIS question


Hi,

    You might want to use Replistor in a case like this and only
have one indexer running, let Replistor manage the copies.

Wes.

-----Original Message-----
From: msftblows@aol.com [mailto:msftblows@aol.com] 
Sent: October 31, 2005 2:28 PM
To: java-user@lucene.apache.org
Subject: Java Indexer + DotLucene + IIS question

Hey-
 
I have the following situation, and I am looking for any suggestions...
 
First, here is my current configuration:
 
1. Java Indexer (windows service) created to index data from a SQL
Server database...3 indexes are created
 
2. DotLucene is used on the front-end to search my index files...which
means I am running on Windows Server 2003 and IIS 6.0 ( sorry guys :-P )
 
3. The index location(s) are read from a config file
 
4. I am currently running this site on one webserver (not a farm)
 
The problem is that I am moving to a webfarm shortly...and I am trying
to figure out a way to replicate the index(s) that I am creating. The
indexer indexes based off a bit field in the database, and a seperate
thread runs in the background to optimize the index(s). 
 
Possible Solutions: 
 
1. If I use so me built-in replication or some other third party
solution across the servers I think (tell me if I am wrong please) I may
have an issue since the indexer compresses and writes and something may
go wrong...
 
2. If I use a shared drive (UNC) to another clustered machine (not one
of the webservers), I need to open ports for IIS and the worker process
to talk to it...not a good idea
 
3. If I store on one web-server and have a shared drive from all other
webservers to that one webserver, then I have a problem if that one
webserver goes down for aby reason
 
4. I don't want to store the indexes in a database
 
5. I can get a Index Server to use (like a state server)...but I would
rather not spend the money :-)
 
6. I can use some remoting solution to bypass the UNC idea...but this
requires a coding change (not a problem if this is the best solution)
 
 
What I am looking for is someone who knows of the best solution...or if
someone implemented something different than the above.
 
Thanks!
-Joe

---------------------------------------------------------------------
To unsubscribe, e-mail: java-user-unsubscribe@lucene.apache.org
For additional commands, e-mail: java-user-help@lucene.apache.org

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