Failed to Submit a Job on the MRS Console

Issue

On the Jobs page of a cluster on the MRS console, the status of a job is Starting but its result is Undefined, as shown in the following figure. The job fails to be submitted to Yarn.

Cause Analysis

The job management function of the console is scheduled and executed by the cluster management module Executor. Therefore, you need to check the Executor to locate the job submission failure. In normal cases, after a job is added, the job management function on the console automatically starts two jobs on Yarn. One job is submitted to the launcher-job queue, which is an auxiliary job queue. The other is submitted to the queue where the job is actually executed, for example, the default queue.

Check the Executor log (/var/log/executor/exe.log on the active master node). It is found that the keytab authentication file fails to be downloaded because the password of the user who submits the Flink job is changed or expires. As a result, the job is not submitted to the launcher-job queue.

Procedure

1. Reset the password of the user who submits the job.

   Log in to Manager, choose System > Permission > User. In the Operation column of the IAM user who submits the job, click More, select Initialize Password, and perform operations as prompted. After the initialization is complete, you need to log in to Manager as this user.

2. Log in to the MRS console and access the MRS cluster. On the Dashboard page, click Synchronize on the right of IAM User Sync.
3. After IAM synchronization is complete, create and submit a job on the Jobs page.