AnsweredAssumed Answered

Mapr 4.2 upgrade - hadoop-0.20.2 references

Question asked by packetboy2000 on Mar 11, 2015
Latest reply on Mar 11, 2015 by packetboy2000
Sorry for all the questions, but still trying to get this upgrade done....8 hours into it and all Tasktrackers fail to start with the same error.  Here are the TT logs:

-----
    java.vm.specification.vendor: Oracle Corporation
    os.name: Linux
    hadoop.id.str: mapr
    tt.tasks.cpu: 20
    sun.jnu.encoding: UTF-8
    tt.tasks.io-spindles: 12.0
    java.library.path: /opt/mapr/hadoop/hadoop-0.20.2/bin/../../hadoop-2.5.1/lib/native:
    hadoop.home.dir: /opt/mapr/hadoop/hadoop-0.20.2/bin/..
    java.specification.name: Java Platform API Specification
    java.class.version: 51.0
    java.net.preferIPv4Stack: true
    sun.management.compiler: HotSpot 64-Bit Tiered Compilers
    hadoop.pid.dir: /opt/mapr/hadoop/hadoop-0.20.2/bin/../pids
    os.version: 2.6.32-358.6.1.el6.x86_64
    user.home: /home/mapr
    
    ------------------------------------------------------------*/
    2015-03-11 01:58:56,087 INFO org.apache.hadoop.mapred.TaskTracker: /tmp is tmpfs. Java Hotspot Instrumentation will be enabled by default
    2015-03-11 01:58:56,089 INFO org.apache.hadoop.mapred.TaskTracker: Cleaning up config files from the job history folder
    2015-03-11 01:58:56,094 ERROR org.apache.hadoop.mapred.TaskTracker: Can not start TaskTracker because java.lang.NumberFormatException: For input string: "4294967296>"
            at java.lang.NumberFormatException.forInputString(NumberFormatException.java:65)
            at java.lang.Long.parseLong(Long.java:441)
            at java.lang.Long.parseLong(Long.java:483)
            at org.apache.hadoop.conf.Configuration.getLong(Configuration.java:1170)
            at org.apache.hadoop.mapred.TaskTracker.<init>(TaskTracker.java:2182)
            at org.apache.hadoop.mapred.TaskTracker.main(TaskTracker.java:5460)
    
    2015-03-11 01:58:56,097 INFO org.apache.hadoop.mapred.TaskTracker: SHUTDOWN_MSG:


So I understand the message, but have no idea which config file it's getting that string from.

Also, why is the java library and hadoop home pointing at hadoop-0.20.2 post upgrade (yes we did run the configure.sh -R script post upgrade)???

Outcomes