Help Center/ MapReduce Service/ Component Operation Guide (LTS)/ Using Flink/ Configuring and Developing a Flink Visualization Job/ Verifying Flink's Job Inspection
Updated on 2024-05-29 GMT+08:00
View PDF

Verifying Flink's Job Inspection

This section applies to MRS 3.3.0 or later.

Scenario

When a large number of Flink jobs are running in a cluster, the FlinkServer web UI provides the Flink job health management function to help you evaluate the health status of each job. You can view the health status of the current job on the page and export the health information of all jobs with just one click. Job statuses are as follows:

  • Healthy: The job is healthy and running properly.
  • Subhealthy:
    • The "ALM-45637 Flink Task Is Continuously Under Back Pressure" alarm is generated. After the alarm is cleared based on the alarm information, the health status automatically recovers to Healthy.
    • The "ALM-45639 Checkpointing of a Flink Job Times Out" alarm is generated. After the alarm is cleared based on the alarm information, the health status automatically recovers to Healthy.
  • Unhealthy:
    • The "ALM-45636 Flink Job Checkpoints Keep Failing" alarm is generated. After the alarm is cleared based on the alarm information, the health status automatically recovers to Healthy.
    • The "ALM-45638 Number of Restarts After Flink Job Failures Exceeds the Threshold" alarm is generated. After the alarm is cleared based on the alarm information and the job is restarted. The health status automatically recovers to Healthy.

Prerequisites

  • The cluster is running properly and the cluster client has been installed.
  • The function of registering jobs with FlinkServer and the job alarming function have been enabled by configuring the Client installation path/Flink/flink/conf/flink-conf.yaml file before submitting a job. The parameter settings are as follows:
    Table 1 Parameters for enabling job registration and job alarming

    Parameter

    Value

    Description

    job.register.enable

    true

    Whether to enable job registration with FlinkServer. Value options are as follows:

    • true: Enable this function.
    • false: Disable this function.

    job.alarm.enable

    true

    Whether to enable job alarming. Value options are as follows:

    • true: Enable this function.
    • false: Disable this function.

    If the function of registering jobs with FlinkServer is not enabled, you cannot start, develop, or stop jobs registered with FlinkServer through the client on the FlinkServer web UI.

  • Ensure that the job is not submitted in session mode and the job name must be specified.

Procedure

  1. Access the Flink web UI by referring to Accessing the Flink Web UI.
  2. Click Job Management. The job management page is displayed.

    • Viewing job health
      On the Job Management page, view the health status of the current job.
      • Empty: The job is not running and has no health status.
      • Green icon: healthy
      • Yellow icon: subhealthy
      • Red icon: unhealthy
    • Exporting all job health reports

      Click Job Health Report. The system automatically exports the health status information of all jobs to the local host, including the job name, health status, submission user, alarm information, configuration information, and start time.

      • If the health score is 0, the job is healthy.
      • If the health score is 1, the job is subhealthy.
      • If the health score is 2, the job is unhealthy.