CODEWITHSUNDEEP
amazon-web-servicesdockeramazon-ecstibco-emsaws-auto-scaling
Ali
Ali

Reputation: 1590

How to drain message from aws ecs docker container which is running behind autoscaling group and connected to jms queue

we are using aws ecs task to run our docker containers. our applications are connecting to TIBCO queues and using AlpakkaJMSListeners to read messages from queues.

We want to enable autoscaling for these containers at ec2 instance level. We have application LB with Target Group defined for some of our services which have http end points.

I know that if we use Target Group with http health check enabled and autoscaling on instances then ecs task will drain requests before stopping container.

Question is how to achieve the same behavior (i.e draining all messages before container is stopped) when our containers are connected to JMS Queues with autoscaling ?

Upvotes: 1

Views: 623

Answers (1)

Ali
Ali

Reputation: 1590

we found a workaround for this one, by default ECS task when triggers TERM signal it waits for 30 sec to let current thread finish their task.

ref - https://docs.aws.amazon.com/AmazonECS/latest/APIReference/API_StopTask.html,
By adding a SIGTERM handler in the listener we can stop consumption of msgs and process inflight messages.

add Terminal signal handling in jms listener like below

Signal.handle(new Signal("TERM"), signal -> {
        log.info("Received termination signal" +signal.getName());
        if ("TERM".equals(signal.getName())) {
            try {
                stopStream();
                log.info("Stopped stream");
            } catch (Exception e) {
                log.error("Exception while handling signal: "+e);
            }
        }
    });

note : in order to receive signal make sure your application is running as a separate java process;

Upvotes: 0

Related Questions