All Products
Search
Document Center

Realtime Compute for Apache Flink:View deployment performance

Last Updated:Nov 28, 2023

You can view the performance of a job that is running, such as the CPU utilization of the JobManager and the CPU utilization, the dynamic memory usage, and the thread usage of TaskManagers that are running. This topic describes how to view the performance of the JobManager and the performance of TaskManagers that are running of a deployment.

Prerequisites

If you want to allow other Alibaba Cloud accounts or RAM users to use your resources in fully managed Flink, you must grant them the required permissions, such as the permissions to view deployment performance. For more information, see Grant permissions to an account.

Background information

You can view the performance of the JobManager and TaskManagers that are running to identify code errors, such as errors of deployment code, slow initialization of classes, and a large number of resources occupied by classes. The following table describes the information about the deployment performance that you can view in the console of fully managed Flink.

Item

Description

Flame Graph

  • CPU: In most cases, functions that are at the top level of the flame graph and occupy a large width consume more CPU resources. These functions may have performance issues.

  • Alloc: You can view the memory usage that is allocated by different types of classes.

  • Lock: You can view the cause of the decline in the performance of a deployment. For example, you can check whether the decline is caused by lock contention or deadlock issues.

  • ITimer: You can view the CPU utilization of all threads at each interval when the time intervals among threads are the same.

Memory Dynamics

You can view the memory changes in different memory regions of the Java Virtual Machine (JVM).

Threads Dynamics

You can view the details about the threads. You can select threads to sample and analyze the details of the threads.

Limits

  • Only Realtime Compute for Apache Flink that uses Ververica Runtime (VVR) 4.0.11 or later allows you to view the deployment performance.

  • You can view only the performance of a job that is in the RUNNING state. You cannot view the performance of a historical job.

Procedure

  1. Log on to the Realtime Compute for Apache Flink console.

  2. On the Fully Managed Flink tab, find the workspace that you want to manage and click Console in the Actions column.

  3. In the left-side navigation pane, choose Deployments.

  4. Click the name of the desired deployment.

  5. Click the Exploration tab.

  6. View the performance of the JobManager and TaskManagers that are running.

    • Performance of the JobManager

      1. On the Running Logs tab, click Job Manager and click Debug.

        Note

        If the keyword "unknown" is displayed on the Flame Graph page, the Non-Java Unknown (native) method is used. If the keyword "unknown" instead of the name of the method in use is displayed on the Flame Graph page, the deployment runs as expected. For more information, see Unknown.

      2. View the performance of the JobManager.

    • Performance of TaskManagers that are running

      1. On the Running Logs tab, click Task Managers and click the value in the Path, ID column.

      2. Click Debug.

      3. View the performance of TaskManagers that are running.