Open danillouz opened 4 years ago
We could update the "status" of the corresponding recording item, so the user knows processing an update failed. A good approach is usually by using a dedicated SNS topic for errors.
Additionally, if we'd want to retry a transcoding job, we should us a Dead Letter Queue (DLQ), but if we want to explore adding this functionality, we should open a separate issue, because it comes with its own challenges.
AWS recently announced Lambda Destinations which seems interesting and convenient to deal with errors.
Currently errors are captured and send to the serverless dashboard, but it would be good to also "flag" the recording item in the DB.