Service check failures

  1. MapReduce service check fails
    Solution:
    1. If the MapReduce service check failed with /user/ambari-qa not found:

      Look for the ambari-qa folder in the DFS user directory. If it does not exist, create it. If this step is skipped, MapReduce service check will fail with the /user/ambari-qa path not found error.

      As root:
      • mkdir <gpfs mount>/user/ambari-qa
      • chown ambari-qa:hadoop /gpfs/hadoopfs/user/ambari-qa
    2. If the MapReduce service check time out or job failed due to permission failure for yarn:

      For Yarn, ensure that the yarn yarn.nodemanager.local-dirs and yarn.nodemanager.log-dirs are writable for the user yarn. If this is not the case, add write permission to the yarn.nodemanager.local-dirs and yarn.nodemanager.log-dirs directories for the user yarn.

  2. What to do when the Accumulo service start or service check fails?

    Solution:

    Note:
    • Ensure that the HDFS and Zookeeper services are running before you proceed.
    • If it is non root environment, run the commands in the workaround steps after logging in with non root user only.
    • If GPFS is unintegrated, remove the tserver.wal.blocksize entry from Accumulo. From Ambari, go to Accumulo > Configs > Custom accumulo-site and remove the tserver.wal.blocksize value and save the configuration.
    • If GPFS is integrated, follow the workaround for tserver.wal.blocksize as mentioned in the FAQ Accumulo Tserver failed to start.

    If the problem still exists, contact IBM® service.

  3. Atlas service check fails.

    Solution:

    1. Restart the Ambari Infra service.
    2. Restart the Hbase service.
    3. Re-run the Atlas service check.
  4. Falcon service check fails.

    Solution:

    This is a known issue with HDP. For information on resolving this issue, go to Falcon Web UI is inaccessible(HTTP 503 error) and Ambari Service Check for Falcon fails: "ERROR: Unable to initialize Falcon Client object".

  5. What to do when the Hive service check fails with the following error: 
    Templeton Smoke Test (ddl cmd): Failed. : {"error":"Unable to access program:
/usr/hdp/${hdp.version}/hive/bin/hcat"}http_code <401>

    Solution:

    HDP is not able to properly parse the ${hdp.version} value. To set the HDP version, execute the following steps:
    1. Get the HDP version for your environment by running the /usr/bin/hdp-select versions command on any Ambari node.
    2. In the Ambari GUI, click HIVE > Configs. Find the templeton.hcat field under the Advanced webhcat-site.
    3. Replace the ${hdp.version} in the templeton.hcat field with the hardcoded hdp.version value found in step a.

      For example, if the value of hdp.version is 2.6.5.0-292, set the templeton.hcat value from /usr/hdp/${hdp.version}/hive/bin/hcat to /usr/hdp/2.6.5.0-292/hive/bin/hcat.

    4. Restart the HIVE service components.
    5. Re-run the HIVE service check.