retry

Attempts to write messages to a child output and if the write fails for any reason the message is retried either until success or, if the retries or max elapsed time fields are non-zero, either is reached.

output:
retry:
output: {}

All messages in Benthos are always retried on an output error, but this would usually involve propagating the error back to the source of the message, whereby it would be reprocessed before reaching the output layer once again.

This output type is useful whenever we wish to avoid reprocessing a message on the event of a failed send. We might, for example, have a dedupe processor that we want to avoid reapplying to the same message more than once in the pipeline.

Rather than retrying the same output you may wish to retry the send using a different output target (a dead letter queue). In which case you should instead use the try output type.

Fields

max_retries

number The maximum number of retries before giving up on the request. If set to zero there is no discrete limit.

backoff

object Control time intervals between retry attempts.

backoff.initial_interval

string The initial period to wait between retry attempts.

backoff.max_interval

string The maximum period to wait between retry attempts.

backoff.max_elapsed_time

string The maximum period to wait before retry attempts are abandoned. If zero then no limit is used.

output

object A child output.