FLEXMEASURES_JOB_TIMEOUT

[nhoening]

The default timeout for jobs is 180 seconds.
Hosts might want to increase this for slower hardware. Such a config setting would do that.

It should probably be either be the ISO duration format we use in other places or be exactly what the RQ docs describe, e.g. '1h', '3m', '5s. My vote is ISO duration, as we have a few other job settings (e.g. TTL) where we also use that.

• document the setting
• use it when new jobs are created (scheduling, forecasting, ingestion)

[Flix6x]

I agree on the ISO duration. I suggest also supporting a mapping to override the default per queue.

So in addition to:

DEFAULT_JOB_TIMEOUT = "PT5M"

with platform default "PT180S",
also support:

JOB_TIMEOUT = dict(
    forecasting="PT5M",
    scheduling="PT10M",
    ingestion="PT180S",  # or just leave out to get the default 180 second, or whatever `DEFAULT_JOB_TIMEOUT` was set to
)

And then the platform default for JOB_TIMEOUT can be an empty dict.