How to view runtime exception logs - Realtime Compute for Apache Flink

When a deployment fails to start or encounters an error at runtime, you can view its exception logs. This topic describes how to view JobManager exceptions, logs of failed TaskManagers, and TaskManager logs for slow checkpoints.

Background

Type	Description
JobManager exceptions	If a deployment fails due to a failover, check for JobManager exceptions. For more information, see View JobManager exceptions. Note If the JobManager fails to start, this failure is not a JobManager exception. In this case, check the startup logs.
Logs of failed TaskManagers	A healthy deployment has no failed TaskManagers. However, for a high-risk deployment, a TaskManager might fail and then be automatically restarted. Check the archived logs of the failed TaskManager to investigate the cause. For more information, see View logs of failed TaskManagers. Note You can view the logs of a failed TaskManager only if log archiving is enabled. The logs are available only for the retention period configured for log archiving.
TaskManager logs for slow checkpoints	If a checkpoint takes a long time to complete at runtime, check the associated TaskManager logs to find the cause. For more information, see View slow checkpoint logs.

Prerequisites

The deployment instance must be in the RUNNING state.

Usage notes

Logs are paginated. Each page displays up to 1 MB of data, containing about 8,000 to 9,000 log entries. This amount is usually sufficient for troubleshooting. If you cannot find the issue on the first page, navigate to other pages.

View JobManager exceptions

Log on to the Realtime Compute for Apache Flink console.
In the Actions column of the target workspace, click Console.
In the navigation pane on the left, choose O&M > Deployments, and then click the name of your target deployment.
On the Logs tab, click the Exceptions tab on the left to view JobManager exceptions.

Note
In the Exception History section, you can view exceptions from the last seven days. You can also filter by exception type to view historical exceptions.

View logs of failed TaskManagers

Log on to the Realtime Compute for Apache Flink console.
In the Actions column of the target workspace, click Console.
In the navigation pane on the left, choose O&M > Deployments, and then click the name of your target deployment.
On the Logs tab, click Logs, and then select the running deployment instance.
Click the Failed Task Managers tab.

View slow checkpoint logs

Log on to the Realtime Compute for Apache Flink console.
In the Actions column of the target workspace, click Console.
In the navigation pane on the left, choose O&M > Deployments, and then click the name of your target deployment.
View the logs for the slow checkpoint.
1. On the Logs tab, click the Checkpoints tab, and then click Checkpoints History.
2. Examine the values in the End to End Duration column to identify the ID of the slow checkpoint.
  
  For example, on the Checkpoints History tab, if the End to End Duration for ID 443 is 3m 24s while others range from 24s to 59s, ID 443 is a slow checkpoint.
3. Click the icon to the left of the slow checkpoint's ID to find the corresponding operator node.
4. Click the icon to the left of the operator node to expand and view the details of each task.
5. Click the ID of the task node corresponding to the slow checkpoint.
  
  This ID is located in the subtask list within the SubTasks section, under Checkpoint Detail. For example, click 0.
6. On the Running Task Managers tab of the page that appears, view the logs for the slow checkpoint.