_Lambda_custom_ec2_cluster.md

Idea

1. set Var for functionality -> get CF output stack (AMI-ID, SQS_QUEUE, S3_bucket)
2. get messages in queue
3. evalute ec2 instances (atm message_count / 5 )
4. check how many instances are running, based on filter (module)
5. calculate required instance ( +2 means -> has to start to more, -2 means stop 2 )
6. if less instances are running start instances
   6.1 start instance
   a. set min & max count to difference of actual running instances and evaluted_number
   b. wait until running
   c. set tag for instance (module)
   6.2 if more instances are running stop instances
   a. get all instances and filter based on tags
   b. stop difference from actual running instances and evaluted_number
7. finish

Problems

• default max ec2 instances are 20
• maybe start bigger instances with parallel translator
• request higher instance limit https://console.aws.amazon.com/support/cases#/create?issueType=service-limit-increase&limitType=service-code-ec2-instances

get_cf_outputs.py

import boto3 


def get_cf_outputs(stack_name):
    cf_conn = boto3.client('cloudformation', region_name='eu-central-1')
    stack = cf_conn.describe_stacks(StackName=stack_name)
    print(stack['Stacks'][0]['Outputs'][0]['OutputValue'])
    # check with index you need couldn´t test it
    return stack['Stacks'][0]['Outputs'][0]['OutputValue']