PrometheusCollector is not filtering application metrics

[jszczepkowski]

If you configure prometheus collector for application metrics, all metrics exposed on the endpoint will be collected. Filtering metrics by name on a base of metrics_config is not implemented.

[jszczepkowski]

According to https://github.com/google/cadvisor/blob/master/docs/application_metrics.md, if you provide the following config:

{
  "endpoint" : "http://localhost:8000/metrics",
  "metrics_config" : [
    {
      "scheduler_binding_latency",
      "scheduler_e2e_scheduling_latency",
      "scheduling_algorithm_latency"
    }
  ]
}

only scheduler_binding_latency, scheduler_e2e_scheduling_latency and scheduling_algorithm_latency metrics should be exposed. However, in practice, all metrics from http://localhost:8000/metrics are exposed.

BTW: the documentation is also broken. The sample config file should be:

{
  "endpoint" : "http://localhost:8080/metrics",
  "metrics_config" : [
      "scheduler_binding_latency",
      "scheduler_e2e_scheduling_latency",
      "scheduling_algorithm_latency"
  ]
}

and the docker label pointing to the file for Prometheus collector should be named io.cadvisor.metric.prometheus.

[DirectXMan12]

I can take a look at fixing this.