Home Forums HDP on Linux – Installation Datanodes not recognized after Ambari upgrade

This topic contains 4 replies, has 2 voices, and was last updated by  tedr 10 months, 2 weeks ago.

  • Creator
    Topic
  • #26899

    We did an upgrade of our Ambari installation from version 1.2.0 to 1.2.3.7. For the upgrade we follwed the instructions according to the manual http://incubator.apache.org/ambari/1.2.3/installing-hadoop-using-ambari/content/ambari-chap7.html

    The upgrade went fine and also the agent upgrade didn’t give any errors. When we log in into the console we see that all the services are up and running. The only thing we don’t see is hostnames. In the host section no hosts are listed. And also in the services section no hostnames are listed (NameNode, JobTracker, Hbase Master) everything is empty. When I click the link to go the namenode UI I get redirected to http://null:50070

    We checked the server logs and we do find some messages:

    Ambari-server log:
    09:49:22,033 WARN ServletHandler:514 – /api/v1/clusters/poc_adm/hosts/null/host_components/NAMENODE
    java.lang.NullPointerException
    09:50:22,111 ERROR ReadHandler:82 – Caught a runtime exception executing a query
    java.lang.NullPointerException

    We do see heartbeats from the datanodes coming into the log.

Viewing 4 replies - 1 through 4 (of 4 total)

You must be logged in to reply to this topic.

  • Author
    Replies
  • #26926

    tedr
    Moderator

    Hi Maik,

    Good to hear that you found the cause! Yeah, the error seems a bit strange as it misdirects you as to where to look for the config file.

    Thanks,
    Ted.

    Collapse
    #26909

    We solved it. The config of the ambari-agent running on the masternode wasn’t copied from conf.save to conf.

    Renaming it back to conf and starting the ambari-agent solved the problem (although the errors still seem strange)….

    Thanks

    Collapse
    #26907

    We did some further investigating today. The agent on the master node refuses to start. When starting it fails with the follwing message:

    INFO 2013-06-04 18:38:33,211 main.py:87 – loglevel=logging.INFO
    WARNING 2013-06-04 18:38:33,211 main.py:124 – No config found, use default
    ERROR 2013-06-04 18:38:33,211 main.py:134 – Ambari prefix dir /tmp/ambari-agent does not exists, can’t continue

    When I check the postgres repository and query the hoststate table I see the following:

    ambari=> select * from hoststate;
    agent_version | available_mem | current_state | health_status | host_name | time_in_state
    ———————+—————+—————+———————————————-+—————————
    {“version”:”1.3.0″} | 14879293 | INIT | {“healthStatus”:”HEALTHY”,”healthReport”:”"} | had-03 | 1370352440205
    {“version”:”1.3.0″} | 14963179 | INIT | {“healthStatus”:”HEALTHY”,”healthReport”:”"} | had-02 | 1370352453793
    {“version”:”1.3.0″} | 15288238 | INIT | {“healthStatus”:”HEALTHY”,”healthReport”:”"} | had-04 | 1370352466228
    {“version”:”v1″} | 65137541 | INIT | {“healthStatus”:”UNKNOWN”,”healthReport”:”"} | had-01 | 1370277913654

    Collapse
    #26904

    tedr
    Moderator

    Hi Maik,

    This is a strange bit. I’ll have to try and replicate it, to see if I can figure out how to recover from it.
    Will keep you posted on my progress.

    Thanks,
    Ted.

    Collapse
Viewing 4 replies - 1 through 4 (of 4 total)