accumulo-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Eric Newton <>
Subject Re: Accumulo and CDH4 -- "no tablet servers"
Date Tue, 09 Apr 2013 02:14:35 GMT
> 2013-04-08 15:18:07,871 [tabletserver.NativeMap] ERROR: Failed to load
native map library blah.... blah... blah

You will need to rebuild the native library for your machine as Josh
suggested.  Remember to push the shared libs out to other servers.

But, this isn't fatal.  Java-only substitutes will be used.

> 2013-04-08 < 2013-04-08>15:18:08,713
[server.Accumulo] WARN : System swappiness setting is greater
than ten (60) which can cause time-sensitive operations to be delayed.
 Accumulo is time sensitive because it needs to maintain distributed lock

Eventually you will want to look into fixing this.  It will be fine for now.

2013-04-08 15:18:08,865 [tabletserver.TabletServer] DEBUG: gc
> ParNew=0.03(+0.00) secs ConcurrentMarkSweep=0.00(+0.00) secs
> freemem=121,318,224(+2,322,040) totalmem=132,055,040

Since you are not using the native map, you may want to increase the size
of your JVM.

> 2013-04-08 15:18:16,536 [tabletserver.TabletServer] INFO : Root tablet
> loaded: !0;!0<<

This is a good sign.  The master is alive and assigning tablets, and the
tablet server(s) are loading them.

> 2013-04-08 15:18:17,131 [tabletserver.TabletServer] DEBUG: UpSess
> 2 in 0.271s, at=[1 1 1.00 1] ft=0.220s(pt=0.001s
> lt=0.217s ct=0.002s)
> 2013-04-08 15:18:17,147 [tabletserver.TabletServer] DEBUG: ScanSess tid
> !0 2 entries in 0.00 secs, nbTimes = [2 2 2.00 1]

Tablets are being read and written to.

Have you tried to bring up the shell and create a table?

> I then add them to general.classpaths in accumulo-site.xml:

OK, this is a little different than HADOOP-2.0, but it seems you muddled
through with the proper paths:

> <property>
>     <name>general.classpaths</name>
>         <!--
>                 Add the following for hadoop-2.0
>         $HADOOP_PREFIX/share/hadoop/common/.*.jar,
>         $HADOOP_PREFIX/share/hadoop/common/lib/.*.jar,
>         $HADOOP_PREFIX/share/hadoop/hdfs/.*.jar,
>         $HADOOP_PREFIX/share/hadoop/mapreduce/.*.jar,
>         $HADOOP_PREFIX/share/hadoop/yarn/.*.jar,
>         -->
>         <value>
>         $ACCUMULO_HOME/server/target/classes/,
>         $ACCUMULO_HOME/core/target/classes/,
>         $ACCUMULO_HOME/start/target/classes/,
>         $ACCUMULO_HOME/fate/target/classes/,
>         $ACCUMULO_HOME/proxy/target/classes/,
>         $ACCUMULO_HOME/examples/target/classes/,
>         $ACCUMULO_HOME/lib/[^.].$ACCUMULO_VERSION.jar,
>         $ACCUMULO_HOME/lib/[^.].*.jar,
>         $ZOOKEEPER_HOME/zookeeper[^.].*.jar,
>         $HADOOP_CONF_DIR,
>         $HADOOP_PREFIX/[^.].*.jar,
>         $HADOOP_PREFIX/lib/[^.].*.jar,
>         $HADOOP_HDFS_HOME/.*.jar,
>         $HADOOP_HDFS_HOME/lib/.*.jar,
>         $HADOOP_MAPREDUCE_HOME/.*.jar,
>         $HADOOP_MAPREDUCE_HOME/lib/.*.jar,
>         $HADOOP_YARN_HOME/.*.jar,
>         $HADOOP_YARN_HOME/lib/.*.jar,
>         </value>
> </property>
> I've verified that these jars are on the classpath (with accumulo
> classpath), but am not sure if I have any I don't need.

I'm not sure either... but they're probably not hurting anything.

> I looked at http://localhost:50095/log and found the following 3 errors,
> in this order:
> org.apache.thrift.transport.TTransportException: Could not create
> ServerSocket on address

This is the file garbage collector.  You may want to check its log.  You
can kill/start the garbage collector without affecting the system much.
On your master, or the machines listed in conf/gc, do this:

$ pkill -f =gc
$ ./bin/

If the monitor page isn't showing anything red or yellow for the garbage
collector, then ignore the messages. Unable to find a listen port

>                 at

Verify you have only one host listed in "conf/masters" or at least not two
hostnames that refer to the same machine.

This is probably a major oversight on my part -- I apologize in advance and
> appreciate any help.

Not at all.  Your system actually looks like it is working.  I'm guessing
you have two entries in your conf/masters file, or those files changed
without a restart in between.

Thanks for taking on the bleeding edge!


View raw message