Hi! I use a

run_failure_sensor

to be notified at every dagster error. I need to be notified quickly so the sensor is set to tick every 30 seconds. It happens in some cases that I run a whole window of time (a hundred partitions) and all these runs fail for the same reason (an unresponsive API, a misconfiguration, etc), so I get a hundred notifications (which is not ideal). Is there a more reasonable way to not get notified on every error? Ideally I want to be notified at the first error (so less than 30 seconds after) while being aware of the following errors. I guess I would have to configure my own

run_failure_sensor

.

Fwiw built-in alerting exists in dagster cloud for most event-based types of errors you might encounter If you wanted to implement this in OSS, you would probably need some sort of custom database that you use to dedupe notifications from incidents