Hi All,
We have a 5 nodes setup running Solr 4.8.1 and we are trying to get the most
out of it by tuning Solr caches.
Following is the output of the script version.sh provided with Tomcat
Server version: Apache Tomcat/7.0.39
Server built: Mar 22 2013 12:37:24
Server number: 7.0.39.0
OS Name: Linux
OS Version: 3.0.76-0.11-default
Architecture: amd64
JVM Version: 1.7.0_21-b11
JVM Vendor: Oracle Corporation
To measure the performance, we are running a simple set of queries using Jmeter
with 25 threads from another host (not a member of our cloud)
We tried to tune the different caches (mostly the documentCache, filterCache
and queryResultCache) to reduce the number of evictions but the cloud
became very unstable at some point. Each server has 92GB of RAM and has 2
collections (1 shard and two replicas) for a total of 6 cores per node.
Each core is around 9.4GB in size according to the Core admin panel.
We have yet to determine where the exact breaking point is.
The two patterns we are seeing are:
- less cache (around 20-30% hit/ratio), poor performance but overall
good stability
- more cache (over 90% hit/ratio), improved performance but almost no
stability.
In that case, we start seeing messages such as "No shards
hosting shard X" or "cancelElection did not find election node to remove"
Anyone, has any advice on what could cause this? I am beginning to suspect the
JVM version, is there any minimal requirements regarding the JVM?
Thanks
