Warning: This section is under construction

🤨 AWS docs use sql syntax highlighting for CloudWatch query syntax... but why?!

Log Groups & Log Streams

Each Lambda function will have a Log Group

If you have a Lambda function named my-function, the log group will look like /aws/lambda/my-function. If the lambda is deployed to Lambda@Edge, the log group will look like /aws/lambda/<REGION>.<FUNCTION_NAME>.

Log Streams

Each log group will have multiple log streams.

Logs Insights

p99, p95, p67, p50,

Note: p99 is your 99th percentile or top 1% of your user group or data pool.

The following CloudWatch query can be visualized as a line, and will depict function duration (or latency) as the p99/95/67/50 for every 10 minutes.

• p50 is your average case; 50% of function invocations
• p99 is your worst case; 1% of function invocations

fields @timestamp, @duration as d
    | filter ispresent(d)
    | sort @timestamp desc
    | stats max(d) as max_MS,
      percentile(d, 99) as p99,
      percentile(d, 95) as p95,
      percentile(d, 67) as p67,
      percentile(d, 50) as p50 by bin(10m)
    | limit 10000

Memory Usage

Note: See Queries for Lambda logs

fields @timestamp, @message
    | sort @timestamp desc
    | limit 50
    | filter @type = "REPORT"
    | stats max(@memorySize / 1000 / 1000) as PROVISIONED_MB,
      min(@maxMemoryUsed / 1000 / 1000) as MIN_MB,
      avg(@maxMemoryUsed / 1000 / 1000) as AVG_MB,
      max(@maxMemoryUsed / 1000 / 1000) as MAX_MB,
      MAX_MB / PROVISIONED_MB * 100 as PERCENT_MAX

Creating a Dashboard with CDK

...🙄...

Alerting

?