hbase-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From st...@apache.org
Subject svn commit: r1211254 - /hbase/trunk/src/docbkx/troubleshooting.xml
Date Wed, 07 Dec 2011 01:18:24 GMT
Author: stack
Date: Wed Dec  7 01:18:24 2011
New Revision: 1211254

URL: http://svn.apache.org/viewvc?rev=1211254&view=rev
Log:
HBASE-4968 Add to troubleshooting workaround for direct buffer oome's.

Modified:
    hbase/trunk/src/docbkx/troubleshooting.xml

Modified: hbase/trunk/src/docbkx/troubleshooting.xml
URL: http://svn.apache.org/viewvc/hbase/trunk/src/docbkx/troubleshooting.xml?rev=1211254&r1=1211253&r2=1211254&view=diff
==============================================================================
--- hbase/trunk/src/docbkx/troubleshooting.xml (original)
+++ hbase/trunk/src/docbkx/troubleshooting.xml Wed Dec  7 01:18:24 2011
@@ -543,6 +543,21 @@ hadoop   17789  155 35.2 9067824 8604364
             </para>
             
        </section>    
+       <section xml:id="trouble.client.oome.directmemory.leak">
+            <title>Client running out of memory though heap size seems to be stable
(but the off-heap/direct heap keeps growing)</title>
+            <para>
+You are likely running into the issue that is described and worked through in
+the mail thread <link xhref="http://search-hadoop.com/m/ubhrX8KvcH/Suspected+memory+leak&amp;subj=Re+Suspected+memory+leak">HBase,
mail # user - Suspected memory leak</link>
+and continued over in <link xhref="http://search-hadoop.com/m/p2Agc1Zy7Va/MaxDirectMemorySize+Was%253A+Suspected+memory+leak&amp;subj=Re+FeedbackRe+Suspected+memory+leak">HBase,
mail # dev - FeedbackRe: Suspected memory leak</link>.
+A workaround is passing your client-side JVM a reasonable value for <code>-XX:MaxDirectMemorySize</code>.
 By default,
+the <varname>MaxDirectMemorySize</varname> is equal to your <code>-Xmx</code>
max heapsize setting (if <code>-Xmx</code> is set).
+Try seting it to something smaller (for example, one user had success setting it to <code>1g</code>
when
+they had a client-side heap of <code>12g</code>).  If you set it too small, it
will bring on <code>FullGCs</code> so keep
+it  a bit hefty.  You want to make this setting client-side only especially if you are running
the new experiemental
+server-side off-heap cache since this feature depends on being able to use big direct buffers
(You may have to keep
+separate client-side and server-side config dirs).
+            </para>
+       </section>    
     </section>
     
     <section xml:id="trouble.mapreduce">



Mime
View raw message