Improve DPA reconcile error messages

[mateusoliveira43]

Problem

It is difficult for users (and developers) to debug OADP errors without more context (studying DPA, for example).

Example

Some months ago, a new user was not able to set custom credential name to BSL (managed by DPA), as DPA would reconcile false with the generic error

Secret "cloud-credentials" not found

After some hours, OADP team asked for user DPA, and the problem was found. There was also VSL defined in DPA (probably copy/paste from example DPA)

oadp-operator/config/samples/oadp_v1alpha1_dpa.yaml

Lines 28 to 33 in 68a1c47

	snapshotLocations:
	- velero:
	provider: aws
	config:
	region: us-west-2
	profile: "default"

This problem could have been resolved immediately if the error message was more informative. Example

VolumeSnapshotLocation <VSL-name> defined in DPA spec.snapshotLocations[<index>] is invalid: 
   create Secret "cloud-credentials" in namespace <namespace-name> or change Secret name in DPA spec.snapshotLocations[<index>].velero.credential.name

Solution

Solution would be to rewrite OADP error messages to be more informative.

A first approach suggestion is to check error messages in tests that are already written, BUT do NOT check error messages.

Example

This test

oadp-operator/controllers/bsl_test.go

Line 57 in 68a1c47

func TestDPAReconciler_ValidateBackupStorageLocations(t *testing.T) {

only checks if error occurred, without testing the error message.

This is very WRONG. Some of these tests are broken, because their error messages do not relate to what is being tested. But they are are not failing in CI, as we check only if an error occurred, and not what error occurred.

Maybe this command can help found other tests like this

grep -Iinr 'wantErr:' . --include=\*_test.go