bug: e2e test will fail randomly

[nic-6443]

Issue description

Because we test UDP route by using kube-dns as backend, we create route in kube-system namespace. But we run e2e in 2 nodes mode, as a result, when the UDP route case is executing, it will affect the accuracy of other cases, because the ingress controller in other cases will also get this UDP route.
This is why e2e failed this time: https://github.com/apache/apisix-ingress-controller/runs/4173189172?check_suite_focus=true

[gxthrj]

Good catch

[gxthrj]

So， I think we should not watch kube-system anywhere.

[nic-6443]

So， I think we should not watch kube-system anywhere.

Should only be watched in UDP case?

[gxthrj]

So， I think we should not watch kube-system anywhere.

Should only be watched in UDP case?

Yes, I think so.

[nic-6443]

I will fix this issue

[tao12345666333]

It seems that the recent e2e is running relatively stable, and related PRs have also been merged, can we close this issue?

[tokers]

@nic-6443