ECS not starting tasks even though there is enough capacity
I've got a task definition that uses 977MB memory and 1024 CPU, using awsvpc networking:
And I have EC2 instances provided by an Auto Scaling group tied to a Capacity Provider, already running 4 tasks across 4 instances:
But when I launch 2 new tasks on top the 4 that are already running, they get stuck in "PROVISIONING" state. Despite what looks like enough memory and CPU to run the new tasks.
What could be going wrong and where can I look further to debug this?
I found the answer.
I was running t3a.small instances, but these have a limit of at most 2 network interfaces including the host. So you can only run one container on them. Limits are documented here: https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/using-eni.html#AvailableIpPerENI
- Access SQL Server with Integrated security from .NET Core app running on Linux container using AWS ECS
- How to deploy single-platform image to ECR after building with --platform flag?
- Is it possible to install Apache superset on a ECS container
- Why does CodePipeline require an `imagedefinitions.json` when ECS task definitions already have this?
- ECS Fargate: return the essential container(s) exit code
- AWS ECS Restart Policy
- EventBridge Scheduler can't update ECS Service
- How to monitor ECS Fargate health check
- Cannot reach ECS instance from different ECS instance
- Bad Gateway ELB ECS Fargate Healthy Target Groups
- AWS advice needed for approach to building a MAUI - API -SqlServer C# app
- How to get the service to which the ECS task belongs
- Is there a setting to get ASP.NET web APIs to accept requests from https in an AWS Fargate container and Route53?
- How to Set an existing ECS cluster to a compute environment in AWS Batch
- Do AWS public facing Load Balancer IPs change?
- How to specify a secret from SecretManager in CloudFormation container definition
- Flask containers on ECS Fargate are completely unresponsive
- Amazon Elastic Container Service and Helm
- Target groups in ECS
- Multiple docker containers in one EC2 instance through AWS ECS
- AWS ECS Service Connect versus Service Discovery
- How to implement instance_requirements from launch template in terraform?
- How to make sure every ECS container I start can always access my network load balancer?
- Port mapping for AWS ECS
- Connecting to AWS DocumentDB from an AWS ECS Container using IAM
- Why do I get an error about executionRoleArn not being specified when it's clearly specified in the file?
- Circular Dependency error with ECS accessing RDS
- AWS Cloudformation Create-Stack Template Error
- Static Files Not Loading in Production in Django-React application
- Is it "bad" to install a system package in a docker entryPoint script?