Fixes #34706 - introduce an option to exclude step from whitelist

[upadhyeammit]

No description provided.

[theforeman-bot]

Issues: #34706

[evgeni]

This doesn't prevent users from whitelisting those steps, right? It just doesn't suggest that it's possible/desireable to do so?

If the above is correct, then the patch should do what you say, but the commit message is misleading ;-)

[jlsherrill]

tested this and this is what i got.

Tasks: TOP => katello:pulp3_content_switchover
(See full trace by running task with --trace)
Checking for valid Katello configuraton.
--------------------------------------------------------------------------------
Scenario [Migration scripts to Satellite 6.10] failed.

The following steps ended up in failing state:

  [content-switchover]

I kinda think we should make this step never skippable. Its asking for too many problems, but i'm fine with whatever support says

[upadhyeammit]

This doesn't prevent users from whitelisting those steps, right? It just doesn't suggest that it's possible/desireable to do so?

If the above is correct, then the patch should do what you say, but the commit message is misleading ;-)

Yeah, I feel there could be still a situation where failures are false positives. In that case users should have option to skip it? Unless we are very sure whatever problems reported by the switchover are always legit.

[evgeni]

I think that there are situations/steps that should not be disable-able, at all.

[upadhyeammit]

This doesn't prevent users from whitelisting those steps, right? It just doesn't suggest that it's possible/desireable to do so?
If the above is correct, then the patch should do what you say, but the commit message is misleading ;-)

Yeah, I feel there could be still a situation where failures are false positives. In that case users should have option to skip it? Unless we are very sure whatever problems reported by the switchover are always legit.

tested this and this is what i got.

Tasks: TOP => katello:pulp3_content_switchover
(See full trace by running task with --trace)
Checking for valid Katello configuraton.
--------------------------------------------------------------------------------
Scenario [Migration scripts to Satellite 6.10] failed.

The following steps ended up in failing state:

  [content-switchover]

I kinda think we should make this step never skippable. Its asking for too many problems, but i'm fine with whatever support says

I like the idea however we need to be 100% sure that whatever failures reported by the content-switchover are always legit. This means there are no situation where we will use whitelist, which I think difficult to agree on?

[jlsherrill]

The switchover step is so vital to be run successfully that i don't think it should be skippable. If a user skips it, they are likely completely destroying all the content views on their satellite and will have to start over from scratch. I'd rather have 100 users debug this failing and work through it than 1 user skipping it and losing everything.

[ekohl]

I think that there are situations/steps that should not be disable-able, at all.

Should his be called "mandatory" instead?

[upadhyeammit]

As per the discussion with Ashish I have modified the message for downstream situation. As example,

The following steps ended up in failing state:

  [pulp-remove]

Resolve the failed steps and rerun
the command.

If the situation persists and, you are unclear what to do next, 
contact Red Hat Technical Support. 

In case the failures are false positives,
use --whitelist="pulp-remove"

[upadhyeammit]

@evgeni added the tests, if you think those can be written in some better way then suggestions are welcome :)

[evgeni]

Can we add a failing step to the tests, that has do_not_whitelist=true, so we can assert that it's not listed in the "you can whitelist things" output?

I think that's the last missing piece before I say ACK ;-)

[upadhyeammit]

Can we add a failing step to the tests, that has do_not_whitelist=true, so we can assert that it's not listed in the "you can whitelist things" output?

I think that's the last missing piece before I say ACK ;-)

done :)

[evgeni]

ACK!