Ruler: register sender errors (thanos_alert_sender_errors_total) metric.

[OGKevin]

Thanos, Prometheus and Golang version used:
Thanos: https://github.com/thanos-io/thanos/releases/tag/v0.10.1

Object Storage Provider:

What happened:
There is a bug in the code here:

thanos/pkg/alert/alert.go

Lines 290 to 318 in 7c02430

	s := &Sender{
	logger: logger,
	alertmanagers: alertmanagers,
	versions: versions,
	sent: prometheus.NewCounterVec(prometheus.CounterOpts{
	Name: "thanos_alert_sender_alerts_sent_total",
	Help: "Total number of alerts sent by alertmanager.",
	}, []string{"alertmanager"}),
	errs: prometheus.NewCounterVec(prometheus.CounterOpts{
	Name: "thanos_alert_sender_errors_total",
	Help: "Total number of errors while sending alerts to alertmanager.",
	}, []string{"alertmanager"}),
	dropped: prometheus.NewCounter(prometheus.CounterOpts{
	Name: "thanos_alert_sender_alerts_dropped_total",
	Help: "Total number of alerts dropped in case of all sends to alertmanagers failed.",
	}),
	latency: prometheus.NewHistogramVec(prometheus.HistogramOpts{
	Name: "thanos_alert_sender_latency_seconds",
	Help: "Latency for sending alert notifications (not including dropped notifications).",
	}, []string{"alertmanager"}),
	}
	if reg != nil {
	reg.MustRegister(s.sent, s.dropped, s.latency)
	}
	return s

The errors metric is not registered in the registry. This means that this metric is missing and therefore not exposed.

What you expected to happen:
This metric to exist.

How to reproduce it (as minimally and precisely as possible):

Full logs to relevant components:

Anything else we need to know:

[bwplotka]

Ups!

[bwplotka]

Ok it's enough, we have too many of such issues. Let's automate this and add some test: #2102