Video streaming devops blog

Updating the ECS AMI's agent

ECS marketing image

The last blog post about downtime was basically due to an old buggy Agent 1.11.1 baked into the ECS AMI which was baked in the ecs-cli orchestration tool.

There are two ways to remedy the situation.

  1. Recommended: Edit your cluster’s Cloud Formation stack with the updated AMI by updating the EcsAmiId parameter, usually found by looking at ecs-cli commits like this one for 2016.09.d.
Editing the ECS AMI in CFN

Then scale up and down. This by default will spawn a new container instance with the updated AMI. Scaling down should kill the old one.

ecs-cli scale --size 4 --capability-iam
ecs-cli scale --size 2 --capability-iam

This will address issues like https://alas.aws.amazon.com/ALAS-2017-783.html.

  1. Update the agent manually from the console via a Update agent button on the console

Do it through the AWS Web console:

List your ECS instances Update the Agent Check the versions
Posted 2016-10-25
Page history

Devops at Spuul. Any tips or suggestions? Reach out!