Take the 2-minute tour ×
Server Fault is a question and answer site for professional system and network administrators. It's 100% free, no registration required.

Today I created an AutoScaling Group with several policies, each with an Alarm associated with a CloudWatch metric. The AutoScaling Group launched two instances, because that the minimum that I required for that group. I turned on my benchmarking tests to hit the new group and their combine average CPUUtilization rose above the thresholds that I had set for the Alarms. I watched the History for each of those Alarms as the states changed from INSUFFICIENT_DATA, then to OK, then finally to ALARM. As expected. Then I waited for the Policies to be executed. The specified 300 seconds before execution came and went. Then I decided to see how long it would take before the Policies executed. It has now been over an hour and a half and those Policies have yet to execute.

I got frustrated and deleted the ASG and all the Policies and Alarms. I decided that perhaps I had made an error in creating the Policies and Alarms. So I recreated them all. Same problem. The Alarms are in ALARM state and never execute the Policies. I can execute the Policies manually without problem, but that defeats the point.

If it makes any difference, I created all these things in the AWS Console. In the past I have used the API to create ASGs and Policies and those groups execute policies just fine.

Is there something I am doing wrong? (Or perhaps AWS services are just having issues.)

UPDATE:

Tried this again today. Decided to use the API instead of the Console to see if there was any difference. The Alarm I am using just updated from OK to ALARM 20 minutes ago.

CPUUtilization >= 35 for 5 minutes
Actions: In ALARM: For group TEST use policy Scale Out API (Add 1 instance)
Namespace: AWS/EC2
Metric Name: CPUUtilization
Dimensions: AutoScalingGroupName = TEST
Statistic: Average
Period: 5 minutes

2013-12-19 15:13 UTC-7 - State update - Alarm updated from OK to ALARM

Current time: 15:43 UTC-7 (30 minutes since alarm)

Action has not triggered. Any idea why?

share|improve this question

1 Answer 1

I asked this question on the Amazon Forums and apparently there is a recent bug in the creation of Alarms that automatically sets the "ActionsEnabled" property to False.

From AWS:

We have identified an issue in the AutoScaling console regarding the binding of
AutoScaling policies to CloudWatch alarms and are working on a fix. We will post
an update to this thread once the fix is rolled out. Thanks for bringing this to
our attention.

The workaround for now:

In the meantime, please try calling the DescribeAlarms CloudWatch API. If the
alarms associated with your policies have ActionsEnabled=false, then this could
cause your policies to not be invoked when the alarm is triggered. Please try
calling the PutMetricAlarm CloudWatch API to update ActionsEnabled=true for the
affected alarms, and that should fix the issues you are experiencing.

I have confirmed the bug and the workaround with my own Alarms through the API.

share|improve this answer

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.