It is a mistake to give java 6gb out of 8gb available. First, when you say 6gb, thats just the heap - the java process will use memory beyond 6gb. That doesn't leave you with hardly any RAM for any other process. And it leaves you almost *nothing* for the filesystem cache. Effectively, you are starving yourself of memory. Lower the heap by a lot - start low- giving java more ram than it needs is not beneficial. So start low and monitor the heap usage using your container, jconsole, or visualgc. If you don't pin the min and max, you will see how much the jvm actually wants but what actually gets allocated (start with a low min for this). Lower your max accordingly.

- Mark


sunnyfr wrote:
Hi, Any idea ?
Sep 25 06:50:41 solr-test jsvc.exec[23286]: Sep 25, 2008 6:50:41 AM
org.apache.solr.common.SolrException log SEVERE: java.io.IOException: Cannot
run program "snapshooter": java.io.IOException: error=12, Cannot allocate
memory My memory for java is JAVA_OPTS="-Xms6000m -Xmx6000m -XX:+UseParallelGC -XX:+AggressiveOpts
-XX:NewRatio=5 -Xloggc:gc.log -XX:+PrintGCDetails -XX:+PrintGCTimeStamps"

My memory is not that bad, I've 8Gb
[EMAIL PROTECTED]:/# free -m
             total       used       free     shared    buffers     cached
Mem:          7998       7095        903          0          6        950
-/+ buffers/cache:       6138       1860
Swap:         2000        295       1705
-------------------------------------------------------------------------------------------------------------
Mem:   8190864k total,  7266832k used,   924032k free,     7280k buffers
Swap:  2048248k total,   302244k used,  1746004k free,   974040k cached



My cache is :
   <filterCache
      class="solr.LRUCache"
      size="16384"
      initialSize="4096"
      autowarmCount="4096"/>

   <!-- queryResultCache caches results of searches - ordered lists of
         document ids (DocList) based on a query, a sort, and the range
         of documents requested.  -->
    <queryResultCache
      class="solr.LRUCache"
      size="16384"
      initialSize="4096"
      autowarmCount="1024"/>

  <!-- documentCache caches Lucene Document objects (the stored fields for
each document).
       Since Lucene internal document ids are transient, this cache will not
be autowarmed.  -->
    <documentCache
      class="solr.LRUCache"
      size="4096"
      initialSize="50000"
      autowarmCount="4096"/>

My index are quite big:
[EMAIL PROTECTED]:/# ls -ll data/solr/book/data
drwxr-xr-x 2 tomcat55 nogroup 8192 Sep 25 10:08 index
drwxr-xr-x 2 root     root    4096 Sep 22 16:20 snapshot.20080922162058



What would you reckon ? Thanks a lot guys,


Reply via email to