Posted On: Oct 23, 2023
Today, HAQM Elastic Container Service (HAQM ECS) enhanced tasks scheduling to make customers’ applications even more resilient to unpredictable load spikes. Now, HAQM ECS will first start a healthy replacement for each unhealthy task, that failed to pass a container or load balancer health check, before terminating it. This enhancement increases the resilience of customers’ applications with no additional efforts or configuration.
HAQM ECS is a fully managed container orchestration service that enables customers to easily run highly secure, reliable, and scalable containerized applications. Customers can define container health or load balancer checks for their applications to let HAQM ECS scheduler know when an unhealthy task should be terminated and replaced with the new one. During the unpredictable load spikes, health checks could provide a false negatives status for busy tasks. This could result for the services to go under capacity, as unhealthy tasks will be terminated prematurely. Now, HAQM ECS will attempt to start a replacement for each unhealthy task and wait until the replacement becomes healthy, before terminating an old task. Customers can control the maximum number of tasks replacements that HAQM ECS can start simultaneously, by configuring the maximumPercent (default 200%) parameter of a service.
The new improvement is available in all AWS regions.
Learn more about HAQM ECS scheduling concepts in this document.