Strange internal telemetry metric naming caused by units

[chenlujjj]

Hi team,

I noticed that the internal telemetry metrics' units were updated from "1" to:

• "{datapoints}"
• "{records}"
• "{spans}"

Related PRs:

• [receiverhelper] Update metric units #10657
• [exporterhelper] Update metric units #10655

This caused the metric name looks strange. For example, if we expose the internal metrics in /metrics endpoint:

> curl http://<collector-host>:8888/metrics

...
# HELP otelcol_receiver_accepted_metric_points__datapoints__total Number of metric points successfully pushed into the pipeline. [alpha]
# TYPE otelcol_receiver_accepted_metric_points__datapoints__total counter
otelcol_receiver_accepted_metric_points__datapoints__total{otel_scope_name="go.opentelemetry.io/collector/receiver/receiverhelper",otel_scope_schema_url="",otel_scope_version="",receiver="prometheus",transport="http"} 151
...

• There are two underscore symbols before and after datapoints
• points and datapoints means the same thing

And if the metrics are ingested into a TSDB like VictoriaMetrics via otlphttp, the metric name is, for example, otelcol_receiver_target_accepted_metric_points_{datapoints} :

Wondering if we can:

• Remove braces of the unit
• Rename the internal metrics, e.g.:
  • otelcol_receiver_accepted_metric_points -> otelcol_receiver_accepted_metric

So the full metric name is otelcol_receiver_accepted_metric_datapoints_total, looks better now.