Fix panic setting ingress status

[dtomcej]

What does this PR do?

Fixes a panic when trying to compare a non-existent ingress status with the expected status

Motivation

Fixes #3490

[errm]

LGTM

Is this fix testable?

[dtomcej]

@errm You mean like can I build an image to test with?

Or can a unit test be written to test this?

Yes to the first, but the 2nd not sure. It relies on the response from the client, not sure how to mock that response in any meaningful manner...

[errm]

Yes I meant a unit test

As this code lives inside of the client it does seem quite hard to test.... Just asking the question though because on the whole adding unit tests to cover regressions like these stop us from re-introducing the same bugs later during refactoring. But having looked closely at this method, I don't see how to do that without refactoring the logical parts into another function...

[dtomcej]

In total agreement.

[timoreimann]

Theoretically, we could unit test the code in question. Practically, we'd likely have to test against a custom, rather complex mock of ours that mimics client-go.

Eventually we'll need integration tests, though client-go's fake implementation could complete the picture. But that'd be a bigger endeavor.

For now: LGTM.

[nmengin]

LGTM 👏