lucene-solr-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Shawn Heisey <>
Subject Solr caches and document deletes
Date Sun, 04 Apr 2010 17:54:15 GMT
I asked this question on Friday evening (US timezone), but nobody's 
responded.  Could be just that it's Easter weekend, but my question was 
a little convoluted, so I'll re-ask it in a simpler way.

If all I'm doing in an index update is deleting documents, it seems that 
it should be possible to keep the existing caches around, and just 
remove references to deleted documents from them, rather than throwing 
them out the window and doing a rewarm.  Is this something that has been 
brought up before, or had an issue filed in Jira?

A slightly related question: If you make a commit with waitSearcher set 
to false, is there a way to then check the status of that commit and 
know when it's completely done?

Another slightly related question:  Is there any way, without changing 
the config file and restarting Solr or reloading the core, to completely 
turn autoCommit off for initial DIH import, then turn it back on for 
updates?  The autoCommit seems to cause issues for my initial import, at 
least on the 1.5-dev that I'm using.  Thinking back to when I was trying 
it with 1.4, the build then worked the same as it does now with maxTime 
and maxDocs both set to 0 - it creates only one .fdt file and multiples 
of all the other files.  Without the explicit settings, it creates lots 
of .fdt files and doesn't delete all the old files when the final 
optimize is done.

I haven't yet tried to look at the source to see if I can get the delete 
behavior I'm after on my own.  I've never been a software developer and 
have very little experience with Java, so jumping into a large work like 
Solr is intimidating.  I'm a system and network administrator, and Perl 
is my drug of choice when I write something, unless it's simple enough 
for shell scripting.  I'm using Perl to write my Solr maintenance system 
right now, which is how I'm coming across issues to discuss here.  I'm 
not using any Solr API in my scripts, just doing GET/POST on the URLs 
with LWP.

If anyone has some pointers about where to start poking around in the 
source, I can take a look.  I'm sure I can get past my unfamiliarity 
with Java's syntax pretty quickly, I just don't really have time to 
figure out the entire program flow on my own.


View raw message