I am currently running a sensor that results in yielding a lot of runs (one for each file found in an s3 bucket). The sensor currently times out due to the amount of files, but never starts any runs even though run requests have been yielded. Is this supposed to happen? My understanding was that once a run request was yielded, it would be launched and ran, but that does not seem to be the case if the sensor times out. It seems to be that if a sensor times out, any yielded runs are "canceled". Is this expected behavior?

Hi Tyler. Yes, by default Dagster has a 60 second timeout for fetching the results from a sensor/schedule. If the sensor/schedule function takes longer than 60s, it will raise an error and not kick off any runs. Ideally, your sensor/schedule wouldn't take longer than 60s to run, but you could change the

DAGSTER_GRPC_TIMEOUT_SECONDS

env var to increase the timeout