Scrape error metric for receiver/sqlquery

[benjaminjb]

Component(s)

receiver/sqlquery

Is your feature request related to a problem? Please describe.

Moving from a non-OTel exporter to this, I found that if the receiver/sqlquery runs into an error, it doesn't have a "scrape error" metric. Should we have one?

What sqlquery does to scrape:
https://github.com/open-telemetry/opentelemetry-collector-contrib/blob/main/internal/sqlquery/scraper.go#L78

Describe the solution you'd like

The postgres_exporter sets a gauge and exports that as a metric:

https://github.com/prometheus-community/postgres_exporter/blob/master/cmd/postgres_exporter/postgres_exporter.go#L680-L699

https://github.com/prometheus-community/postgres_exporter/blob/master/cmd/postgres_exporter/postgres_exporter.go#L542

(It is kind of open-ended: any number/type of errors gets reported as a scrape error "y/n" output, but it's useful to know if the scraping itself has run into an error)

Describe alternatives you've considered

Possibly there's something in the OTel internal metrics that should be leveraged for this sort of warning?

https://opentelemetry.io/docs/collector/internal-telemetry/

Additional context

No response

[github-actions]

Pinging code owners:

• receiver/sqlquery: @dmitryax @crobert-1

See Adding Labels via Comments if you do not have permissions to add labels yourself.

[benjaminjb]

Hi @iblancasa, I'm curious: is this a reasonable feature request that you'd like to talk about more or should I look at the OTel internal metrics (which does have a otelcol_scraper_errored_metric_points_total counter) as a replacement for this?

[benjaminjb]

Oh, and for an alternate idea: sql_exporter has a scrape error metric, but it's (a) a counter and (b) has a reset timer (optional).