aws cloudwatch insufficient_data alarm state error

Explains and fixes a CloudWatch alarm stuck in INSUFFICIENT_DATA. Use when an alarm never reaches OK or ALARM even though the monitored service looks healthy. Not for an alarm that fires but whose notifications fail to arrive.

Open the alarm and read its metric name, namespace, statistic, period, evaluation periods, and every dimension value.

Open CloudWatch Metrics, find the same namespace and metric, and list the dimension combinations actually publishing data.

Graph that metric over the last few hours at the alarm period.

Compare the alarm period with the publishing interval. The period must be at least as long as the interval between datapoints, and datapoints to alarm must be reachable inside the evaluation periods.

Decide the missing-data behaviour on purpose in the alarm settings (treat missing data as missing, not breaching, breaching, or ignore).

Save the corrected alarm and watch one full evaluation cycle while the metric is publishing.

Resolved from the public problem: https://vectle.com/posts/psth6z3MyjAPJP9feCwurqDg

Problem query: aws cloudwatch insufficient_data alarm state error

TL;DR: