I use Solr 3.6.0 and all selects are blocked during commits.
This appears to be contrary to all documentation I have seen, quoting from
the "Solr Enterprise.." book:
"No matter how long a commit or optimize command takes, Solr still
executes searches concurrently—there is no read lock."
I have Master replicating to Slave, once/minute.
Indexing happens to the Master, at a rate of few docs/about every 30 secs
Selects are run against Slave.
This is the pattern from the Slave log:
Oct 30, 2012 12:33:23 AM org.apache.solr.core.SolrDeletionPolicy
updateCommits
INFO: newest commit = 1349195567630
Oct 30, 2012 12:33:42 AM org.apache.solr.core.SolrCore execute
INFO: [core3] webapp=/solr path=/select
During the 19 seconds that you see between the 2 lines, the /select is
blocked, until the commit is done.
This has nothing to do with the jvm, I'm monitoring the memory and GC stats
with jConsole and log, and they are fine.
I played with all settings imaginable: commitWithin, commit=true,
useColdSearcher, autoWarming settings from 0 on-nothing helps.
As soon as I suspend replication (command=disablepoll), everything becomes
fast.
As soon as I enable it - it pretty much becomes useless.
The problem does not seem to be related to replication, because querying
Master directly exibits the same problem-cannot select while commit is
running.
The environment is: Solr 3.6.0, RHEL Lunux 5.3.2, 64-bit, 96G RAM, 6 CPU
cores, java 1.6.0_24, ~70 million docs.
Thanks a lot for your help.
P.S. this message was originally posted to an old thread discussing the same
problem, I'm reposting new per Erick Erickson's request.
--
View this message in context:
http://lucene.472066.n3.nabble.com/solr-selects-blocking-on-commit-tp4017525.html
Sent from the Solr - User mailing list archive at Nabble.com.
