1.4.0 breaks CloudWatch CNI Metrics

[guillaume-eb]

I don't see any error in the cni-metrics-helper pod logs, but Cloudwatch stopped gettings metrics at 10:15 in the screenshot below

I0424 11:29:32.263814       8 wesley-k8s-metrics.go:127] Collecting metrics ...
I0424 11:29:32.263833       8 discovery.go:118] GetCNIPods start...
I0424 11:29:32.263862       8 discovery.go:127] GetCNIPods discovered[aws-node-fg2d5...
I0424 11:29:32.264068       8 metrics.go:404] Grab/Aggregate metrics from aws-node-fg2d5
I0424 11:29:32.264073       8 cni_metrics.go:99] Grabbing metrics from CNI aws-node-fg2d5

[mogren]

Hey @guillaume-eb, thanks for reporting this!

I closed the old PR for #113 and opened #413 instead. We changed the names of the prometheus endpoints in v1.4.0 (PR #348), but since the code for the CNI metrics agent was just in a branch in Liwen's fork, we missed updating it.

I'll cherry-pick the config back to the release-1.4 branch once the #413 has been reviewed and merged.