run_dpa_daily.py

[nrawolk]

The code run_dpa_daily.py still refers to the PSMC model. If this is not needed, perhaps it should be removed from GitHub.

[jeanconn]

From the commit 188f5f469ab message, this is not "converted" yet from the psmc check setup. So run_dpa_daily.py has the not-yet-implemented feature of actually working by calling the dpa model.

[nrawolk]

Ah- Thanks. I am curious if I should convert this for dea_check, or just drop it from the package because I doubt we are going to run it on a daily basis.

Nance

[jeanconn]

We've got the psmc_check set up to run daily to give us reasonably quick email warnings on bad predictions vs telemetry. No mail is sent if everything is within tolerances. That might not be a bad model to follow for all of the models.

[jeanconn]

Granted, the warnings are sent from the task_schedule that calls the application and not directly from the application, so if ACIS wanted to do this independently, you'd need that whole setup too.

[nrawolk]

We check the model validations each week as part of our load review. I wonder if that would be sufficient? Thanks for your help.

[jeanconn]

In practice, our daily auto-validation checking has been more of a check that the commanded states are reasonable than a verification of the model.
Also, for verification of loads, if the model and telemetry differ in the past, you may not care as you're going to be using telemetry directly as a propagation point. You may, however, want to know if there's weirdness or a serious problem before the FOT has generated weekly products.

[taldcroft]

@nrawolk - I think it's OK to drop the daily checking from other ACIS models except for the original PSMC that we run (for cmd states validation as @jeanconn says).