too much commit time at Master/Slave

[zeous <ze...@gmail.com>]

Hi~

i have some questions about commit time at Solr Master/Slave 
my system is that total data is about 1,000,000 and solr version is 1.2 
2.4 CPU and centos, 1G memory, JDK 1.5 

For the test, 
master server does only indexing that 1 data indexs per 10 seconds 
and snapshots by postCommit in solrconfig.xml
slaver server snappullers and snapinstallers per 1 minute by crontab

test case is

1) only indexing
when i don't send slave server select query, master sever does indexing and
slave server is sent snapshots.
slave server's snapinstaller time is 0~1 seconds. i thinks that it is
normal.

2) indexing and select
master server does indexing and slaver server is send snapshots
per 2 second, i send slaver server a different query request.
slaver server's snapinstaller time is 6~9 seconds. 
(query request time is 0~1 seconds.)

At this time i stop send query request but slaver server's snapinstaller
time is the same.
i think that  slaver server's snapinstaller time reduce like master server.
(i watch slaver server during 1 hour, but slaver server don't reduce this
time)

after i restart jetty server, this time is 0~1 second. 

what's problems?
-- 
View this message in context: http://www.nabble.com/too-much-commit-time-at-Master-Slave-tf4749942.html#a13582146
Sent from the Solr - User mailing list archive at Nabble.com.