Skip to main content

Job Diagnosis

Job diagnosis may encounter various problems during execution. To help solve these problems efficiently, Kylin provides a task diagnostic function, which can package related problems' log information into a compressed package for operations or Apache Community supports to analyze problems and ascertain the cause.

View Job Execution Log On Web UI​

You can view the job execution log in the Monitor -> Batch Job interface. As shown below, you can click the Open Job Steps button at the position 1, and then click the Log Output button in the job details to view the first and the last 100 lines of job execution log in a popup window. You can download the full log by clicking the link download the full log at the position 3.

Tip: If there are multiple steps in a job, you can view the execution log for each step.

Job Log

Generate Job Diagnostic Package By Scripts​

In FusionInsight, you need to execute the command source /opt/hadoopclient/bigdata_env first. The hadoopclient is a variable.

You can execute $KYLIN_HOME/bin/diag.sh -job <jobid> to generate the job diagnostic package and \<jobid> need to be replaced with the actual job ID number. You can view the job ID number in the Monitor -> Batch Job interface. You can also click the icon in the position 1 as shown picture below to expand the specified job details that is in the position 2 on the right.

Job ID

The diagnostic package is stored by default in the $KYLIN_HOME/diag_dump/ directory.

After extracting the diagnostic package, you can view the diagnostic package information in the appropriate directory or file.

  • /conf: configuration information under the $KYLIN_HOME/conf directory.
  • /hadoop_conf: configuration information under the $KYLIN_HOME/hadoop_conf directory.
  • /metadata: metadata files.
  • /logs: specifies the logs generated during the execution of job.
  • /spark_logs: specifies all spark executor logs generated during job execution.
  • /system_metrics: specifies the system metrics during the execution of job.
  • /audit_log: specifies the audit logs during the execution of job.
  • /job_tmp: specifies the temporary files, logs and optimization suggestions log of job.
  • /yarn_application_log: specifies the logs of yarn application of job.
  • /client: operating system resources occupied information, hadoop version and kerberos information.
  • /monitor_metrics：The node monitoring log of the specified task.
  • /write_hadoop_conf：$KYLIN_HOME/write_hadoop_conf, Hadoop configuration of the build cluster. This directory will not be available when Read/Write separation deployment is not configured.
  • file catalog_info: directory structure of install package.
  • file commit_SHA1: git-commit version.
  • file hadoop_env: hadoop environment information.
  • file info: license, package and hostname.
  • file kylin_env: Kylin version, operating system information, Java related information, git-commit information.

Tips: If you want to exclude metadata files, please specify -includeMeta false.

Generate Job Diagnostic Package in Web UI​

Job diagnostic package includes all diagnostic information of a specific job, and users can generate job diagnostic package on Web UI by following the following steps:

Generate Job Diagnostic Package in Web UI

  1. In the action bar of a job on the Jobs List page, click the Download Job Diagnostic Package button in Actions.
  2. Select Server.
  3. Click Generate and Download button: After the diagnostic package is generated, the downloads will be triggered automatically. If the diagnostic package generation fails, you can view the details of the failure on the interface.