Rudder

Hiding CR configured with an incorrect PI seems to be a bad idea (it was done so when we tried to make more robust CR deployment facing unexpected behaviour.

So, what is the good thing to do ? Have a log message + disactivate the CR (it seems to be the bare minimum). Remove the inexisting PI ? We can't know anything about the PI, we just know it's ID and has the PI is missing, we are not able to find what PT it was derived from, etc).

Any more ideas ?

I think we should display the CR if it's there, hiding broken things is a bad idea, IMO.

Ideally, we should display it in the table of CRs but put in the "PI" column something like "Policy Instance not found!", and make the effective status "disabled".

Applied in changeset ae759cc38fdc4cc1578ec4f5aba6817965c2e050.

Now, configuration rules with errors are displayed in the configuration rule grid as follow:

You can see the detail of a cr in error and change the part in error to repair the cr.
A configuration rule in error is disabled and will have to be re-enabled after being repaired.

Sounds correct