Support-bundle cli logs failures for ceph even though rook is not enabled

[xavpaice]

Description

When generating a support bundle via CLI, the command logs errors/failures trying to collect and analyze rook ceph even though rook is not enabled in a cluster.

Reproduction Steps

1. Deploy a kURL cluster without rook
2. run kubectl support-bundle https://kots.io

Expected Behavior

We should not attempt to collect and analyze data for addons that are not enabled.

Customer Impact

This is confusing for the customer. It adds noise and shows a failure that is a false negative.

Possible Solution

Edit the Ceph and Longhorn collectors to first determine if that component is installed. If not, do not attempt any collection, simply log a message to the effect that the collector was skipped.

Workaround

Environment

Additional Information

[xavpaice]

Shortcut #44437

[xavpaice]

This is more generic - e.g. when any collector is included that is not able to run because the component is not installed. This one is from running the Longhorn collector without it installed:

 * failed to run collector: Longhorn: list nodes.longhorn.io: the server could not find the requested resource (get nodes.longhorn.io)

Errors like this should not be printed on stdout unless --debug is specified. The information should simply be logged in the bundle output.

The change that would close this bug report:

• [ ] output from the Ceph collector logs failures to collect in logs, rather than stdout unless --debug is supplied
• [ ] output from the Longhorn collector logs failures to collect in logs, rather than stdout unless --debug is supplied
• [ ] The analyzers for Ceph, Longhorn and Weave do not report errors if there is no data to analyze, merely log information to that effect