Ah, this is a classic container networking issue. The problem occurs because port 8080 is the default port that Prefect workers use for their API server, and you can't have two processes binding to the same port on the same network namespace. Here are three ways you can resolve this: 1. Use Different Ports (Recommended): Configure each worker to use a different port by setting the

SERVER_PORT

environment variable. For example:

yaml
   # Task definition excerpt
   containerDefinitions:
     - name: worker1
       environment:
         - name: SERVER_PORT
           value: "8080"
     - name: worker2
       environment:
         - name: SERVER_PORT
           value: "8081"

2. Use

awsvpc

Network Mode: Configure your ECS task to use

awsvpc

network mode. This gives each container its own network interface, allowing them to use the same port without conflict. Update your task definition:

json
   {
     "networkMode": "awsvpc",
     "containerDefinitions": [
       // your container definitions here
     ]
   }

3. Use Dynamic Port Mapping: If using

bridge

network mode, let Docker assign dynamic host ports:

json
   {
     "networkMode": "bridge",
     "containerDefinitions": [
       {
         "portMappings": [
           {
             "containerPort": 8080,
             "protocol": "tcp"
           }
         ]
       }
     ]
   }

The first option (using different ports) is typically the most straightforward and gives you the most control. Just make sure your task definition's security group allows traffic on both ports if you need external access. For more details on ECS-specific configurations, you can refer to the Prefect ECS guide. Would you like me to elaborate on any of these solutions or provide more specific configuration examples for your use case?