Open TheRealGoku opened 3 years ago
You are completely right @TheRealGoku, this was a while back the way to add the Spot interruption notification handler. Things have changed. A while back ago (about 2 years ago) we added the right handling mechanism into the ECS-Optimized AMI image. Now just by adding the ECS_ENABLE_SPOT_INSTANCE_DRAINING variable and setting it to true as documented here you should be able to get the right design.
We will update and remove this entry. I'll close the issue once that this old example gets removed.
Thats true :)
https://github.com/awslabs/ec2-spot-labs/blob/master/ecs-ec2-spot-fleet/ecs-ec2-spot-fleet.yaml#L442
If customer uses spot-instance-termination-notice-handler.sh script in their environment due the change curl version output format, the code will always execute else statement and thereby will keep all the on boarded registered container instances into draining state for lifetime despite whether the instance is on demand or spot