Skip to main content

Restart strategy in SQL and JAR applications

Learn how Aiven for Apache Flink® applications uses restart strategies to recover from job failures, ensuring high availability and fault tolerance for your distributed applications.

About restart strategy

A restart strategy is a set of rules that Apache Flink® adheres to when dealing with Flink job failures. These strategies enable the automatic restart of a failed Flink job under specific conditions and parameters, which is crucial for high availability and fault tolerance in distributed and scalable systems. Aiven for Apache Flink® supports restart strategies for both JAR and SQL applications.

Default restart strategy for JAR and SQL applications

Aiven for Apache Flink® uses the exponential-delay strategy as the default restart mechanism for both JAR and SQL applications. This strategy incrementally increases the delay between restarts, reaching a configurable maximum. After reaching the maximum, the delay remains constant. The strategy resets the exponential delay after a period of successful restarts, preventing it from staying at the maximum indefinitely. This default strategy is integrated into the Aiven for Apache Flink cluster configuration and automatically applies to JAR and SQL applications.

View the default strategy

You can view the default restart strategy configurations for your Aiven for Apache Flink cluster in the Apache Flink Dashboard. Follow these steps to view the current settings:

  1. Access the Aiven Console and select the Aiven for Apache Flink service.
  2. From the Connection information section on the overview page, copy the Service URI and paste it into your web browser's address bar.
  3. When prompted, log in using the User and Password credentials specified in the Connection information section.
  4. Once in the Apache Flink Dashboard, click the Job Manager from the menu.
  5. Switch to the Configuration tab.
  6. Review the configurations and parameters related to the restart strategy.

Disable default restart strategy

While Aiven for Apache Flink® typically recommends following the default restart strategy for high availability and fault tolerance, there might be scenarios, especially during testing or debugging, where disabling automatic restarts might be beneficial.

JAR applications

For JAR applications, you cannot disable the default restart strategy in Aiven for Apache Flink® through configuration files. Instead, directly modify the code of your Jar application to achieve this.

StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();
env.setRestartStrategy(RestartStrategies.noRestart());

This code sets the restart strategy to None, preventing any restart attempts in case of failures.

SQL applications

For SQL applications, you have a simplified approach to restart strategies. You can enable or disable restarts on failure during application deployment, providing a straightforward way to manage applications without complex configurations. For more information, see Create an SQL application.

Key considerations when disabling default restarts

Before disabling the default restart strategy for your applications, consider the following:

  • Persistent failures: Disabling restarts means that if a Flink Job fails, it will not attempt to recover it, leading to permanent job failure.
  • Testing and debugging: Disabling is beneficial when identifying issues in the application code, as it prevents the masking of errors through automatic restarts.
  • External factors: Jobs can fail due to external factors, such as infrastructure changes or maintenance activities. If you disable restarts, your Flink jobs will become vulnerable to failures.
  • Operational risks: In production environments, it is generally advisable to use the default restart strategy to ensure high availability and fault tolerance.