Exploring high-availability and fault-tolerance

The cluster would not be reliable if it wouldn't be fault tolerant. Kops intents to make it so, but we're going to validate that anyways.

Let's retrieve the list of worker node instances:

aws ec2 \
    describe-instances | jq -r \
    ".Reservations[].Instances[] \
    | select(.SecurityGroups[]\
    .GroupName==\"nodes.$NAME\")\
    .InstanceId"  

We used aws ec2 describe-instances to retrieve all the instances (five in total). The output was sent to jq, which filtered them by the security group dedicated to worker nodes.

The output is as follows:

i-063fabc7ad5935db5
i-04d32c91cfc084369  

We'll terminate one of the worker nodes. To do that, we'll pick a random one, and retrieve its ID.

INSTANCE_ID=$(aws ec2 \
    describe-instances | jq -r \
    ".Reservations[].Instances[] \ 
 ...