如何在树莓派3 B+的Raspbian系统上运行AWS ecs-cli

Hey there! Let's break down how to get the AWS ECS CLI up and running on your Raspberry Pi 3 B+ (Raspbian OS), handle ECS container scaling, and keep things within the AWS Free Tier while you plan your Lambda integration.

Quick heads-up: The command you tried uses the amd64 (x86_64) binary, which won't work on your Pi's ARM-based architecture. Here's the corrected setup:

1. First confirm your Pi's architecture (Raspberry Pi 3 B+ typically uses armhf or arm64):

   dpkg --print-architecture
   

2. Download the matching ARM binary:

   • For armhf:

     sudo curl -o /usr/local/bin/ecs-cli https://s3.amazonaws.com/amazon-ecs-cli/ecs-cli-linux-armhf-latest
     

   • For arm64:

     sudo curl -o /usr/local/bin/ecs-cli https://s3.amazonaws.com/amazon-ecs-cli/ecs-cli-linux-arm64-latest
     

3. Make the binary executable:

   sudo chmod +x /usr/local/bin/ecs-cli
   

4. Verify the installation worked:

   ecs-cli --version
   

Once installed, here are the core commands to scale your ECS services:

• Check your cluster's current service and task status:

  ecs-cli ps --cluster your-cluster-name
  

• Scale a specific service to your desired number of tasks (replace N with your target count):

  ecs-cli scale --cluster your-cluster-name --service your-service-name --desired-count N
  

Note: If you're using Fargate, ensure your task definition has sufficient resources allocated; for EC2-based clusters, confirm your EC2 instances have enough CPU/memory to support the scaled tasks.

Since you're aiming to stay within the Free Tier, here are actionable steps to cut CloudWatch costs:

• Disable non-critical ECS metrics: In your ECS service's monitoring settings, uncheck custom metrics you don't need (stick to the default basic metrics initially)
• Set log expiration policies: Configure CloudWatch Log Groups to auto-delete logs after 30 days (or shorter) to avoid long-term storage fees
• Avoid unnecessary API calls: Use ecs-cli's local output to check service status instead of frequent CloudWatch API requests
• Use CloudWatch Free Tier limits strategically: The Free Tier includes 10 custom metrics, 1 million API requests, and 5GB of log storage per month—stick to these thresholds

When you're ready to move to Lambda for scaling, here's a quick preview:

• You can use Lambda to trigger ECS scaling events (e.g., based on time schedules or simple custom triggers) without relying on constant CloudWatch polling
• For cost control, use Lambda's Free Tier (1 million requests/month, 400,000 GB-seconds of compute) to handle scaling logic without extra charges

内容的提问来源于stack exchange，提问作者Otto Mixa