Hi,

We are using Elasticsearch-0.90.5 in our search application. We have 3 
master nodes and 3 data nodes each with heap size of 4gb running in JDK 
1.6. We have our Node Client running in our APP_SERVER. Elasticsearch 
master gets HIGH_CPU frequently at least for every 20 days and it is not 
coming down until that node is restarted. When i see thread dump, nothing 
peculiar except GC. Number of GC threads is almost 20. *GC MarkSweep 
Frequency* and *GC ParNew Frequency* is very very less when compared to 
other nodes. 

When i restart the master, system load becomes normal. But our APP_SERVER 
gets HIGH_CPU. With our monitoring tool, we could detect some packets were 
dropped due to network buffer overflow. When we restart APP_SERVER, all 
things become normal. 

For every 20 days, we have to restart our master and then our APP_SERVERS. 
This has become our routine. I have attached Elasticsearch Masters thread 
dump here <https://gist.github.com/anonymous/10015420>

We couldn't analyze why we are getting HIGH_CPU repeatedly. Any help in 
resolving this issue will be appreciated. 

Thanks in advance

Ranjith Venkatesan


PS : Elasticsearch version and JDK version cant be updated immediately in 
our case.

-- 
You received this message because you are subscribed to the Google Groups 
"elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To view this discussion on the web visit 
https://groups.google.com/d/msgid/elasticsearch/aaa14b57-3253-42dc-9a67-3378bb892a3f%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Reply via email to