This sounds very complicated for only 30K documents. Put them all on one
server, give it enough memory so that the index can all be in file buffers. If
there is a disaster, reindex everything. That should only take a few minutes.
And don't optimize.
wunder
On Sep 8, 2013, at 3:01 PM, atuldj.jadhav wrote:
Hi Team,Need your suggestions/views on the approach I have in place for SOLR
availability and recovery.
I am running *SOLR 3.5* and have around *30k* document's indexed in my SOLR
core. I have configured SOLR to hold *5k * documents in each segment at a
time.I periodically commit optimize my SOLR index.
I have delta indexing in place to index new documents in SOLR, /very rarely
/ I face index corruption issue, to fix this issue I have *checkindex -fix*
job in place as well.However sometime this job can delete the corrupt
segment! (meaning loss of 5K documents, till I full Re-index SOLR.)
_*I have few follow up questions on this case.*_
1. How can I avoid loss of 5K documents (checkindex -fix), shall I reduce
number of documents per segments count? is there an alternate solution?
2. If I start taking periodic backup (snapshot) of entire index, shall I
just replace my data/index folder from the backup folder in case corruption
is found? Is this a good implementation?
3. Any other good solution, suggestion to have maximum index availability
all the time?
Thanks in advance for giving your time.
Atul
--
View this message in context:
http://lucene.472066.n3.nabble.com/SOLR-index-Recovery-availability-tp4088782.html
Sent from the Solr - User mailing list archive at Nabble.com.
--
Walter Underwood
wun...@wunderwood.org
