Introduce metric for per-exporter pressure on prometheus itself

[harmw]

Is it possible to expose some kind of metric to point out the amount of stress a certain metrics endpoint poses on the prometheus machine itself? We have several exporters in our platform, just no simple way of figuring out which one of those is stressing prometheus the most. Some more insights in this process would be great.

[brian-brazil]

What you're mostly looking for is the number of samples that an exporter is producing. Usually it's just a handful of metrics in one exporter. http://www.robustperception.io/which-are-my-biggest-metrics/

[fabxc]

You can also do something along the lines of:

count by(instance) ({__name__=~".+"})

That can be expensive though.

[harmw]

Thanks, that looks useful.

However:

Error executing query: query timed out in expression evaluation

I'll see if I can do something to get the information out of prom.

[brian-brazil]

scrape_samples_scraped does this now.

[lock]

This thread has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.