Help Center/ MapReduce Service/ FAQs/ Component Configurations/ How Do I Handle the Kudu Service Exceptions Generated During Cluster Creation?
Updated on 2024-08-16 GMT+08:00

How Do I Handle the Kudu Service Exceptions Generated During Cluster Creation?

Viewing the Kudu Service Exception Logs

  1. Log in to the MRS console.
  2. Click the name of the cluster.
  3. On the page displayed, choose Components > Kudu > Instances and locate the IP address of the abnormal instance.

    If the Components tab is unavailable, complete IAM user synchronization first. (On the Dashboard page, click Synchronize on the right side of IAM User Sync to synchronize IAM users.)

  4. Log in to the node where the abnormal instance resides, and view the Kudu log.
    cd /var/log/Bigdata/Kudu
    [root@node-master1AERu kudu]# ls
    healthchecklog  runninglog  startlog
    You can find the Kudu health check logs in the healthchecklog directory, the startup logs in the startlog directory, and the Kudu process run logs in the runninglog directory.
    [root@node-master1AERu logs]# pwd
    /var/log/Bigdata/kudu/runninglog/master/logs
    [root@node-master1AERu logs]# ls -al
    kudu-master.ERROR   kudu-master.INFO   kudu-master.WARNING

    Run logs are classified into three types: ERROR, INFO, and WARNING. Each type of run logs is recorded in the corresponding file. You can run the cat command to view run logs of each type.

Handling Kudu Service Exceptions

The /var/log/Bigdata/kudu/runninglog/master/logs/kudu-master.INFO file contains the following error information:
"Unable to init master catalog manager: not found: Unable to initialize catalog manager: Failed to initialize sys tables async: Unable to load consensus metadata for tablet 0000000000000000000000: xxx"

If this exception occurs when the Kudu service is installed for the first time, the KuduMaster service is not started. Data inconsistency causes the startup failure. To solve the problem, perform the following steps to clear the data directories and restart the Kudu service. If the Kudu service is not installed for the first time, clearing the data directories will cause data loss. In this case, migrate data and clear the data directory.

  1. Search for the data directories fs_data_dir, fs_wal_dir, and fs_meta_dir.

    find /opt -name master.gflagfile

    cat /opt/Bigdata/FusionInsight_Kudu_*/*_KuduMaster/etc/master.gflagfile | grep fs_

  2. On the cluster details page, choose Components > Kudu and click Stop Service.
  3. Clear the Kudu data directories on all KuduMaster and KuduTserver nodes. The following command uses two data disks as an example.

    rm -Rvf /srv/Bigdata/data1/kudu, rm -Rvf /srv/Bigdata/data2/kudu

  4. On the cluster details page, choose Components > Kudu and choose More > Restart Service.
  5. Check the Kudu service status and logs.