0

Say I'm looking to reduce average memory use across an auto scaling group - depending on the size of the group, I want to launch a varying number of servers.

I've created a cloudwatch alarm that fires as soon as memory is above some level, and spins up one server. Is there a way to configure the alarm to keep spinning up servers (to some maximum number) until the average memory drops?

blueberryfields
  • 45,910
  • 28
  • 89
  • 168
  • 1
    Not sure what you are trying to achieve and what is the problem? CloudWatch Alarm will continue firing off as long as your memory doesn't drop below alarm level. After an auto scaling action takes place (ex. server added), there's a "cooldown period" during which no Auto Scaling action will take place. Once cooldown period is over another Auto Scaling action will take place if your metric is still in alarm state. You can also configure Auto Scaling to start 1,2,3,4 etc servers at once or go by current group size percentage size depending on how much alarm threshold was breached. – Michal Gasek Nov 24 '16 at 08:12
  • 1
    That is exactly how it works, what is the issue? – Mahdi Nov 24 '16 at 14:09
  • Oh interesting - might be that the boot period + cooling off period are too long for me, so should start more servers at once per @MichalGasek. Thank you! – blueberryfields Nov 24 '16 at 15:09

0 Answers0