If the Lambda worker errors on a changeset more than X times, make a record of what failed then remove it from Kinesis so processing can move forward. Since Lambda is stateless, this will require using Redis or Postgres.
My instance is stuck on a changeset that encounters a bug in one of the metric calculations. The problem is Lambda just keeps pulling and retrying that same changeset over and over.
If the Lambda worker errors on a changeset more than X times, make a record of what failed then remove it from Kinesis so processing can move forward. Since Lambda is stateless, this will require using Redis or Postgres.
My instance is stuck on a changeset that encounters a bug in one of the metric calculations. The problem is Lambda just keeps pulling and retrying that same changeset over and over.