Pre-render prometheus labels

[binarin]

Proposed Changes

Pre-render labels for prometheus exported. Character escaping is not needed, as AMQP object names are quite restricted in what characters they can have.

Doing it this way removes not only CPU overhead, but also a lot of garbage collection overhead (as for every output line labels were converted from a proplist to a list of records before rendering).

This makes per-object metrics twice as fast, time to produce ~630k metrics (10k queues/10k consumer/10k producers) went from ~6 to ~3 seconds from this change.

Depends on deadtrickster/prometheus.erl#137

This also pulls an performance optimization deadtrickster/prometheus.erl#136 - across all those changes on my test machine I saw scraping time dropping 40 times for the aforementioned 630k metrics.

Testing was performed by running 2 instances of perftest with the following arguments (just with a slightly modified queue pattern each time):

--queue-pattern 'perf-test-%d' \
  --queue-pattern-from 1 --queue-pattern-to 5000 \
  --producers 1 --consumers 1 --rate 1

2 instances because perftest was running out of memory, and running two instances was easier for me than learning how to tune java memory settings.

The timing of following urls are of special interest for testing:

http://localhost:15692/metrics/per-object
http://localhost:15692/metrics/detailed?family=queue_coarse_metrics&family=queue_consumer_count
http://localhost:15692/metrics/detailed?family=queue_metrics

Types of Changes

• Bug fix (non-breaking change which fixes issue #NNNN)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause an observable behavior change in existing systems)
• Documentation improvements (corrections, new content, etc)
• Cosmetic change (whitespace, formatting, etc)
• Build system and/or CI

Checklist

• I have read the CONTRIBUTING.md document
• I have signed the CA (see https://cla.pivotal.io/sign/rabbitmq)
• I have added tests that prove my fix is effective or that my feature works
• All tests pass locally with my changes
• If relevant, I have added necessary documentation to https://github.com/rabbitmq/rabbitmq-website
• If relevant, I have added this change to the first version(s) in release-notes that I expect to introduce it

[michaelklishin]

Oh, that's a nice improvement indeed!

[HoloRin]

@pjk25 self reminder to fixup the bazel commit

[HoloRin]

@binarin git_repository refers to an external repository that has it's own bazel files. github_bazel_erlang_lib is a helper we have to import a non-bazel repo into bazel-land.

[HoloRin]

@Mergifyio rebase

[mergify]

rebase

✅ Branch has been successfully rebased

[HoloRin]

@binarin I think this is looking good now

[michaelklishin]

I can confirm 20-50% reduction in response latency (the number of samples is small but the results are consistent).

[michaelklishin]

Backported to v3.9.x manually for 3.9.9.