Closed jlduran closed 3 years ago
Are we sure we want the following behaviour in the case of error:
If the file does not exist, do not display a diagnostic message or modify the exit status to reflect an error.
If someone uses these instructions in a script, yes.
I'm fine with this if it doesn't change the indempotent nature of the step and doesn't introduce new cases that we havent considered, or cant/dont handle
Well, this step is not idempotent. It will remove the keys, and if the server is not restarted (i.e. the keys are not regenerated) they won't/can't be removed again because they don't exist.
The script will however not fail if the file(s) doesn't(don't) exist with rm -f
, I guess that's why Debian and RedHat guides rm -f
(although, I believe the root account on those systems alias rm -i
to rm
, this way it won't prompt for confirmation).
Again, I thought this was going to be an easy "ship it", if you're having doubts, I don't mind leaving it just rm
, and maybe add a warning that if you plan on using these instructions in a scripted installation/provisioning/configuration orchestration, use rm -f
.
Not having doubts. Just wanted to make sure I understood the changed behaviour expectations.
By indempotent i just mean same resulting state (no keys) even with multiple invocations, not that there's no side effect in the first case.
I think the rm-f is fine
It will avoid prompting the user for confirmation, as many times users alias
rm -i
torm
. It will also not fail if the files do not exist.