Add Prometheus Remote Write Exporter supporting Cortex - conversion and export for Histogram OTLP metrics

[huyan0]

This PR is part of a series of PRs implementing a Prometheus remote write exporter supporting Cortex.
See related PR #1577

Description: This PR adds export support for histogram and summary metrics for Prometheus remote write integrated backends, such as Cortex. The exporter drops non-cumulative monotonic, histogram, and summary OTLP metrics.

Please note this metrics exporter does not support Prometheus default labels such as job or instance labels. An issue addressing Prometheus default labels will be filed later. Another related feature is to derive labels from a Resource. This functionality already exists in the Go SDK and will be implemented in another PR. This feature could allow users to specify which attributes they want to add as labels.

Link to tracking Issue: #1150
Related issues are:
Metrics aggregation proposal: #1422
Prometheus exporter not functional: #1255
Related spec discussion: #731

Documentation:

• readme.md with sample configuration, file structure, and assumptions

• Design documentation: PR Add Cortex and Prometheus Remote Write exporter design #1464

cc: @huyan0 @alolita @jmacd @bogdandrutu

[codecov]

Codecov Report

Merging #1643 into master will increase coverage by 0.03%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master    #1643      +/-   ##
==========================================
+ Coverage   92.37%   92.40%   +0.03%     
==========================================
  Files         265      265              
  Lines       19909    19952      +43     
==========================================
+ Hits        18390    18437      +47     
+ Misses       1090     1088       -2     
+ Partials      429      427       -2     

Impacted Files	Coverage Δ
exporter/prometheusremotewriteexporter/helper.go	100.00% <ø> (ø)
exporter/prometheusremotewriteexporter/exporter.go	95.94% <100.00%> (+5.37%)	⬆️
service/defaultcomponents/defaults.go	85.45% <100.00%> (+0.26%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 09d5ee7...ebb3f11. Read the comment docs.

[bogdandrutu]

Please split this into 2 PRs the first being the Histogram. We will not support Summary for few weeks in the new otlp and that will get delayed.

[bogdandrutu]

Please increase the coverage :)

[huyan0]

Please increase the coverage :)

Will do. Also rebasing : )

[huyan0]

Got 100% diff coverage now

[bogdandrutu]

A lot of unnecessary changes can you please fix them?

[huyan0]

Addressed some comments. Really appreciate the review. Will start working on #1681

[vishiy]

@huyan0 @bogdandrutu @EdZou @ercl - I see the prometheus histograms represented as OTEL histograms, during which the values are becoming delta ( see design here - https://github.com/open-telemetry/opentelemetry-collector/blob/main/receiver/prometheusreceiver/DESIGN.md#histogram ). I see the same issue with summary (for _sum & _count) and also counters. Is there a way to configure the receiver to NOT do the delta/diff from first scrape ? If not, how to get the original values as scraped for these metrics to be able to ingest to prometheus thru the prometheus exporter?