PROF-8520: Add DNS events to timeline

[szegedi]

What does this PR do?

Gathers DNS performance events and adds them to the timeline.

Motivation

2023 Q4 Continuous Profiler OKR 5

Additional Notes

This is best reviewed commit by commit, or at least by reviewing the first commit separately, then the others together. The reason for this is that the first commit refactors the timeline profiler from only gathering GC events to having an architecture for gathering multiple types of events. This already pays off in spades – you can see how "Add DNS" is a pretty small commit after the refactor – but it will make adding Net events similarly trivial too.

Security

Datadog employees:

• If this PR touches code that signs or publishes builds or packages, or handles credentials of any kind, I've requested a review from @DataDog/security-design-and-guidance.
• This PR doesn't touch any of that.

[github-actions]

Overall package size

Self size: 5.6 MB
Deduped: 61.6 MB
No deduping: 62.35 MB

Dependency sizes

name	version	self size	total size
@datadog/native-iast-taint-tracking	1.6.4	16.43 MB	16.44 MB
@datadog/native-appsec	5.0.0	15.16 MB	15.17 MB
@datadog/pprof	4.0.1	9.32 MB	10.16 MB
protobufjs	7.2.4	2.74 MB	6.52 MB
@datadog/native-iast-rewriter	2.2.1	2.27 MB	2.36 MB
@opentelemetry/core	1.14.0	872.87 kB	1.47 MB
@datadog/native-metrics	2.0.0	898.77 kB	1.3 MB
@opentelemetry/api	1.4.1	780.32 kB	780.32 kB
import-in-the-middle	1.4.2	41.4 kB	704.79 kB
pprof-format	2.0.7	588.12 kB	588.12 kB
msgpack-lite	0.1.26	201.16 kB	281.59 kB
opentracing	0.14.7	194.81 kB	194.81 kB
semver	7.5.4	93.4 kB	123.8 kB
@datadog/sketches-js	2.1.0	109.9 kB	109.9 kB
lodash.sortby	4.7.0	75.76 kB	75.76 kB
lru-cache	7.14.0	74.95 kB	74.95 kB
ipaddr.js	2.1.0	60.23 kB	60.23 kB
ignore	5.2.4	51.22 kB	51.22 kB
int64-buffer	0.1.10	49.18 kB	49.18 kB
istanbul-lib-coverage	3.2.0	29.34 kB	29.34 kB
lodash.uniq	4.5.0	25.01 kB	25.01 kB
limiter	1.1.5	23.17 kB	23.17 kB
dc-polyfill	0.1.2	22.77 kB	22.77 kB
retry	0.13.1	18.85 kB	18.85 kB
lodash.kebabcase	4.1.1	17.75 kB	17.75 kB
node-abort-controller	3.1.1	16.89 kB	16.89 kB
lodash.pick	4.4.0	16.33 kB	16.33 kB
jest-docblock	29.7.0	8.99 kB	12.76 kB
crypto-randomuuid	1.0.0	11.18 kB	11.18 kB
path-to-regexp	0.1.7	6.78 kB	6.78 kB
koalas	1.0.2	6.47 kB	6.47 kB
methods	1.1.2	5.29 kB	5.29 kB
module-details-from-path	1.0.3	4.47 kB	4.47 kB

_{🤖 This report was automatically generated by heaviest-objects-in-the-universe}

[codecov]

Codecov Report

Attention: 82 lines in your changes are missing coverage. Please review.

Comparison is base (cdebb59) 85.06% compared to head (c7e4d31) 84.87%.

Files	Patch %	Lines
...ackages/dd-trace/src/profiling/profilers/events.js	3.52%	82 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #3822      +/-   ##
==========================================
- Coverage   85.06%   84.87%   -0.20%     
==========================================
  Files         228      230       +2     
  Lines        9363     9579     +216     
  Branches       33       33              
==========================================
+ Hits         7965     8130     +165     
- Misses       1398     1449      +51     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[pr-commenter]

Benchmarks

Benchmark execution time: 2023-11-29 11:25:23

Comparing candidate commit c7e4d31 in PR branch szegedi/PROF-8520-timeline-dns with baseline commit cdebb59 in branch master.

Found 0 performance improvements and 0 performance regressions! Performance is the same for 521 metrics, 11 unstable metrics.

[nsavoire]

🤓 nitpick: ‏ Using the timestamps of the earliest and latest events to determine profile start and duration might be inaccurate.

[szegedi]

Yeah, although I'm unsure what I could use instead of it…

[nsavoire]

Maybe take start/stop timestamps in start / profile functions.
Or have Profiler class take these timestamps and pass them along, that would have the nice side effect of all profiles having the same start and duration.

[szegedi]

that sounds like a great idea; I filed a JIRA issue in the timelines epic to track this.

[nsavoire]

LGTM !
Thanks for splitting the PR into easily reviewable commits 😀

[szegedi]

Hey @nsavoire, thank you for the review, but I have updated the code in the meantime :-(

I had to remove one commit (the one that removed locations from the GC events) – turns out the profiler backend ignores samples that are entirely without a location. I did rework it though to create one synthetic dummy location per profile and use it for all samples. I have a change in the pipeline for UI that makes these invisible (just as they are for .Net profiles).

I also experimented with various ways to visualize DNS events and avoid overlap between events. The easiest I could work out so far is to scatter them into multiple virtual "DNS" threads as needed, so I added that functionality too. A small test app that does:

const dns = require('node:dns')

require('dd-trace').init().profilerStarted().then(() => {
  dns.lookupService('13.224.103.60', 80, () => {})
  dns.lookup('example.org', () => {})
  dns.lookup('datadoghq.com', () => {})
  dns.resolve4('datadoghq.com', () => {})
  dns.lookup('dfslkgsjkrtgrdg.com', () => {})
  setTimeout(() => {
    dns.lookup('example.com', () => {})
  }, 200)
})

will produce a profile that renders like this (also thanks to some backend changes I made to group these DNS threads in a single group):

If you can re-review the changes in the last 4 commits, it'd be appreciated; I'll open a new PR for any subsequent work.

[nsavoire]

🚀