CODEWITHSUNDEEP
amazon-web-servicesaws-lambdaamazon-kinesis
meet_learner
meet_learner

Reputation: 1

Cloudwatch metric parameters in AWS lambda

Iam trying to understand cloudwatch metric for aws lambda.For troubleshooting iterator age related issue ,we may need to investigate following parameters.

  1. memory issues
  2. not enough kinesis shrads
  3. invocation errors
  4. insufficient batch size

Of these , memory issue can be seen from the cloudwatch insight? But how could we identify the issue is because of other parameters? Say how can I identify iterator age is increased and due to not enough shards, or due to batch size or due to invocation errors. Is there any logs to know these? Please help

Upvotes: 0

Views: 303

Answers (1)

uri parush
uri parush

Reputation: 11

Each one of those options requires different metrics and logs.

  1. Check if my lambda is failing - you can check the lambda metrics tab to see if you get errors (lambda finish without success ) in that case all the records from the shard that fail return back to the shard for retry.

  2. A more advanced issue might be having a specific shard that causes the iterator age to grow, in this case, you might need to enable shard metric under kinesis configuration (Enhanced [shard-level] metrics) to find the specific shard that might cause the issue.

In both cases check your lambda logs and metric under CloudWatch to understand if you have errors or timeouts.

  1. You might process your data slower than pushing it in, compare Get records - sum (Count) vs Incoming data - sum (Count) under kinesis metrics in cloudwatch, if you are not processing the data fast enough adding more shards can help (more lambda instances to process data)

As a disclaimer, I work for Lumigo that provides monitoring and debugging solutions for serverless applications. So we can easily address your lambda issues, we will give you full visibility on invocations errors and memory issues, no more searching in logs :-)

Upvotes: 1

Related Questions