Different retry options based on failure type

[mfateev]

Currently both workflow and activity retry options do not differentiate between failure types besides allowing to not retry configured list of failures.

The proposal is to allow specifying different retry options for different failure types. For example intermittent errors can be retried immediately, but some errors that are not intermittent (like NPE) and require human intervention can be retried with much higher intervals.

[ronnie660]

upvoting this ticket! This feature will be very useful for activities that are hitting external APIs and needs to backoff for some period of time on specific error type.

[utkarshpaliwal9]

Just came here to create a similar issue. Also, it should be possible for the workflow to not even retry in the cases of certain types of failure (which are not transient). This error mapping with retry policy can be provided by temporal or be taken as input from the user.

[dhoepelman]

Also, it should be possible for the workflow to not even retry in the cases of certain types of failure (which are not transient).

@utkarshpaliwal9 This is possible, you can mark a failure type string as not retryable or throw a non retryable application failure.

[MatanYadaev]

Is there any workaround?