Configuration for no-retry http response codes on endpoint pipes
Planned
Some endpoints needlessly "hammer" service APIs with retries.
The API responds with a validation error (4xx). In the case that the payload is wrong (and will never validate unless changed in sesam) it does not make sense to send it over and over again. We have examples where this puts unnecessary stress on receiving server.
In other cases validation means that data is missing in receiving system and that sending the same payload again makes sense.
It would be good to be able to distinguish the former and e.g. immediately make entity dead based on http response code.
-
Official comment
Indeed. It it unnecessary to retry requests that returned 4xx as these are client side problems. We'll look into how we can handle this.
Comment actions -
We now support customizing retry behaviour based on HTTP response codes, see https://docs.sesam.io/hub/documentation/service-configuration/systems/configuration-systems-url.html#url-system-retry-strategy. This configuration should work for your case:
"retry_stategy": {
"400-499": {
"max_retries_per_entity": 0,
"batch_retries": 0
}
}
Please sign in to leave a comment.
Comments
2 comments