scrape: Prometheus server support exemplar in native histogram

[fatsheep9146]

After prometheus/client_model#80 is merged
we should mirror the change and support parsing those exemplars and ingesting them, as @beorn7 suggested

prometheus/client_model#80 (comment)

[beorn7]

Thank you.

The next step would be to actually use the new protobuf field.

Currently, we parse the exemplars from the classic buckets, even if we are ingesting native histograms. I guess the behavior should be that we don't do that if there is at least one exemplar in the new dedicated field. @fatsheep9146 would you be interested in implementing that, too?

[fatsheep9146]

Thank you.

The next step would be to actually use the new protobuf field.

Currently, we parse the exemplars from the classic buckets, even if we are ingesting native histograms. I guess the behavior should be that we don't do that if there is at least one exemplar in the new dedicated field. @fatsheep9146 would you be interested in implementing that, too?

Yes, of course. I will also implement that too! @beorn7

[Nexucis]

Is it safe to include this PR in the next release (v2.50) @beorn7 @fatsheep9146 ? Just asking if I can simply rebase the release branch based on main or I will need to cherry pick a bug fixed that has been merged this morning on main to avoid to include this one.

[beorn7]

I think it is safe to include this PR. The code added here is simply not exercised anywhere.