Troubleshooting guide for common issues encountered when using Amazon ECS with Harness Database DevOps, including connectivity problems, container provisioning errors, and rollback execution issues.
Check CloudWatch logs for the runner task. Verify the HARNESS_DELEGATE_TOKEN and HARNESS_MANAGER_ENDPOINT environment variables are correct in the task definition. Ensure the security group allows outbound HTTPS traffic to Harness Manager.
Verify the task execution role has the AmazonECSTaskExecutionRolePolicy attached and includes permissions for logs:CreateLogGroup, logs:CreateLogStream, logs:PutLogEvents. Ensure the role can pull images from ECR if using private images.
Verify the security group has a self-referencing inbound rule for TCP port 20001. Check that both the runner task and build tasks use the same security group. Port 20001 is required for lite-engine communication between the runner and step containers.
Split steps into multiple step groups. Each step group creates a separate ECS task, avoiding the 10-container limit per task. Avoid combining Apply and Rollback steps in the same step group.
Configure a shared log group in the pipeline task configuration to reduce CreateLogGroup calls. AWS CloudWatch has a 5 TPS limit for CreateLogGroup, which can be exceeded when creating many log groups concurrently.
Verify security group inbound rules allow traffic from the ECS task security group to the Aurora database. Confirm the ECS tasks are in the same VPC as Aurora or have proper routing configured. Test connectivity using telnet or nc from a test container.
Separate rollback into its own step group. Ensure proper failure strategy configuration in the pipeline. Rollback steps configured within the same step group as Apply steps may not execute correctly due to ECS task lifecycle constraints.
Go to Set Up Database DevOps on ECS Fargate for complete ECS Fargate configuration instructions.
Go to Rollback for database schemas to configure rollback strategies.
Last updated
Was this helpful?
Was this helpful?
