redhat_subscription: stop manual unsubscribing on unregistration

[ptoscano]

SUMMARY

Unregistering a system also drops all the resources for it automatically, so there is no need to manually unsubscribing (which actually means removing all the subscriptions).

In addition to that, newer versions of subscription-manager drop all the support for entitlements, so the "remove" subcommand (used by unsubscribe()) does not exist anymore, and thus the unregistration fails with those versions.

This fixes the registration on EL 10 systems, and Fedora 41 and greater.

ISSUE TYPE

• Bugfix Pull Request

COMPONENT NAME

redhat_subscription

[ansibullbot]

cc @cnsnyder
click here for bot help

[richm]

Should the code be conditional then? e.g.

                if subman_version < X.Y:
                    rhsm.unsubscribe()
                rhsm.unregister()

?

That is - if I am a customer using this module to manage an EL9 and an EL10 system, how should I code my play to work correctly on both of those systems?

Am I correct in assuming that it is correct to do the rhsm.unsubscribe() when using earlier versions of subman, but incorrect when using later versions?

[ptoscano]

Should the code be conditional then?

No, it shouldn't be needed, that's why I simply dropped the call.

[russoz]

Should the code be conditional then?

No, it shouldn't be needed, that's why I simply dropped the call.

Hi @ptoscano would care to expand? The concern expressed by @richm seems legit: AFAICT this will break backward compatibility for people using the module in RHEL9 or earlier. But I have not tested that.

[ptoscano]

Should the code be conditional then?

No, it shouldn't be needed, that's why I simply dropped the call.

Hi @ptoscano would care to expand? The concern expressed by @richm seems legit: AFAICT this will break backward compatibility for people using the module in RHEL9 or earlier.

No, it will not break anything. Unregistering removes all the bits of a system from the registration server, so removing some manually first is a redundant step.

This is even in very old RHEL's documentation:
https://docs.redhat.com/en/documentation/red_hat_enterprise_linux/5/html/deployment_guide/un-registering#un-registering

The only thing required to unregister a machine is to run the unregister command. This removes the system's entry from the subscription service, removes any subscriptions, and, locally, deletes its identity and subscription certificates.

Emphasis mine on "removes any subscriptions", which is what subscription-manager removes --all (removed by this patch) does.

[russoz]

Thanks for clarifying. In that case, it LGTM.

[patchback]

Backport to stable-9: 💚 backport PR created

✅ Backport PR branch: patchback/backports/stable-9/bcc92e8aac4397117e14386ff462c7399bd7db6e/pr-9578

Backported as #9589

🤖 @patchback
I'm built with octomachinery and
my source is open — https://github.com/sanitizers/patchback-github-app.

[felixfontein]

@ptoscano thanks for your contribution!
@richm @russoz thanks for reviewing!

[patchback]

Backport to stable-10: 💚 backport PR created

✅ Backport PR branch: patchback/backports/stable-10/bcc92e8aac4397117e14386ff462c7399bd7db6e/pr-9578

Backported as #9590

🤖 @patchback
I'm built with octomachinery and
my source is open — https://github.com/sanitizers/patchback-github-app.