[netpath] Add rate limit to dynamic paths

[pimlu]

What does this PR do?

This PR adds a rate limit to the dynamic paths collector. It offers a guarantee that no more than N traceroutes will be run per host, per minute. This limit is currently set to 150 by default.

It is configurable by setting network_path.collector.pathtest_max_per_minute. If it's set to 0, it is disabled.

Additionally, there is network_path.collector.pathtest_max_burst_duration which is not intended to be changed by the customer. It determines the math for how large the rate limiter's burst is. It is 30s by default, so for example, if pathtest_max_per_minute is 150, then the burst is 75 (which 30 seconds' worth of path tests, half a minute).

Motivation

Increase confidence we can deploy dynamic paths to more clusters.

Describe how you validated your changes

PathTestStore has some new tests:

go test -timeout 30s -tags linux,linux_bpf,npm,process,test github.com/DataDog/datadog-agent/comp/networkpath/npcollector/npcollectorimpl/pathteststore

Also, npCollectorImpl tests should still pass:

go test -timeout 30s -tags linux,linux_bpf,npm,process,test github.com/DataDog/datadog-agent/comp/networkpath/npcollector/npcollectorimpl

Possible Drawbacks / Trade-offs

I'm not a fan of the way I init the PathTestStore in start(). It needs to happen after statsd.Client is ready though, so idk if there's a better way.

Additional Notes

[agent-platform-auto-pr]

Test changes on VM

Use this command from test-infra-definitions to manually test this PR changes on a VM:

inv aws.create-vm --pipeline-id=55445725 --os-family=ubuntu

Note: This applies to commit acb0a73

[agent-platform-auto-pr]

Static quality checks ✅

Please find below the results from static quality gates

Info

Result	Quality gate	On disk size	On disk size limit	On wire size	On wire size limit
✅	static_quality_gate_agent_deb_amd64	845.09MiB	858.45MiB	203.57MiB	214.3MiB
✅	static_quality_gate_docker_agent_amd64	929.37MiB	942.69MiB	310.73MiB	321.56MiB

[agent-platform-auto-pr]

Uncompressed package size comparison

Comparison with ancestor ba9f4f1915e0337e5798541bbc533c74932241e0

Diff per package

package	diff	status	size	ancestor	threshold
datadog-heroku-agent-amd64-deb	0.01MB	⚠️	445.91MB	445.91MB	0.50MB
datadog-iot-agent-aarch64-rpm	0.01MB	⚠️	82.72MB	82.72MB	0.50MB
datadog-iot-agent-arm64-deb	0.01MB	⚠️	82.65MB	82.65MB	0.50MB
datadog-agent-amd64-deb	0.00MB	✅	873.71MB	873.70MB	0.50MB
datadog-agent-x86_64-rpm	0.00MB	✅	883.45MB	883.44MB	0.50MB
datadog-agent-x86_64-suse	0.00MB	✅	883.45MB	883.44MB	0.50MB
datadog-agent-arm64-deb	0.00MB	✅	861.58MB	861.58MB	0.50MB
datadog-agent-aarch64-rpm	0.00MB	✅	871.30MB	871.30MB	0.50MB
datadog-iot-agent-amd64-deb	0.00MB	✅	86.38MB	86.38MB	0.50MB
datadog-iot-agent-x86_64-rpm	0.00MB	✅	86.45MB	86.45MB	0.50MB
datadog-iot-agent-x86_64-suse	0.00MB	✅	86.45MB	86.45MB	0.50MB
datadog-dogstatsd-x86_64-rpm	0.00MB	✅	59.18MB	59.18MB	0.50MB
datadog-dogstatsd-x86_64-suse	0.00MB	✅	59.18MB	59.18MB	0.50MB
datadog-dogstatsd-amd64-deb	0.00MB	✅	59.10MB	59.10MB	0.50MB
datadog-dogstatsd-arm64-deb	0.00MB	✅	56.57MB	56.57MB	0.50MB

Decision

⚠️ Warning

[cit-pr-commenter]

Regression Detector

Regression Detector Results

Metrics dashboard
Target profiles
Run ID: abaaa5c2-279d-48c9-827f-24a8a22e96ba

Baseline: ba9f4f1
Comparison: acb0a73
Diff

Optimization Goals: ✅ No significant changes detected

Fine details of change detection per experiment

perf	experiment	goal	Δ mean %	Δ mean % CI	trials	links
➖	quality_gate_logs	% cpu utilization	+3.41	[+0.28, +6.54]	1	Logs
➖	quality_gate_idle_all_features	memory utilization	+0.70	[+0.63, +0.76]	1	Logs bounds checks dashboard
➖	quality_gate_idle	memory utilization	+0.35	[+0.31, +0.39]	1	Logs bounds checks dashboard
➖	uds_dogstatsd_to_api_cpu	% cpu utilization	+0.35	[-0.52, +1.22]	1	Logs
➖	tcp_syslog_to_blackhole	ingress throughput	+0.07	[-0.00, +0.14]	1	Logs
➖	file_to_blackhole_0ms_latency_http1	egress throughput	+0.05	[-0.81, +0.91]	1	Logs
➖	file_to_blackhole_0ms_latency	egress throughput	+0.02	[-0.75, +0.79]	1	Logs
➖	file_to_blackhole_500ms_latency	egress throughput	+0.02	[-0.76, +0.80]	1	Logs
➖	uds_dogstatsd_to_api	ingress throughput	+0.02	[-0.27, +0.31]	1	Logs
➖	tcp_dd_logs_filter_exclude	ingress throughput	-0.01	[-0.03, +0.01]	1	Logs
➖	file_to_blackhole_300ms_latency	egress throughput	-0.02	[-0.66, +0.63]	1	Logs
➖	file_to_blackhole_1000ms_latency_linear_load	egress throughput	-0.04	[-0.51, +0.43]	1	Logs
➖	file_to_blackhole_0ms_latency_http2	egress throughput	-0.06	[-0.93, +0.82]	1	Logs
➖	file_to_blackhole_100ms_latency	egress throughput	-0.09	[-0.83, +0.65]	1	Logs
➖	file_to_blackhole_1000ms_latency	egress throughput	-0.09	[-0.91, +0.72]	1	Logs
➖	file_tree	memory utilization	-0.29	[-0.36, -0.21]	1	Logs

Bounds Checks: ✅ Passed

perf	experiment	bounds_check_name	replicates_passed	links
✅	file_to_blackhole_0ms_latency	lost_bytes	10/10
✅	file_to_blackhole_0ms_latency	memory_usage	10/10
✅	file_to_blackhole_0ms_latency_http1	lost_bytes	10/10
✅	file_to_blackhole_0ms_latency_http1	memory_usage	10/10
✅	file_to_blackhole_0ms_latency_http2	lost_bytes	10/10
✅	file_to_blackhole_0ms_latency_http2	memory_usage	10/10
✅	file_to_blackhole_1000ms_latency	memory_usage	10/10
✅	file_to_blackhole_1000ms_latency_linear_load	memory_usage	10/10
✅	file_to_blackhole_100ms_latency	lost_bytes	10/10
✅	file_to_blackhole_100ms_latency	memory_usage	10/10
✅	file_to_blackhole_300ms_latency	lost_bytes	10/10
✅	file_to_blackhole_300ms_latency	memory_usage	10/10
✅	file_to_blackhole_500ms_latency	lost_bytes	10/10
✅	file_to_blackhole_500ms_latency	memory_usage	10/10
✅	quality_gate_idle	intake_connections	10/10	bounds checks dashboard
✅	quality_gate_idle	memory_usage	10/10	bounds checks dashboard
✅	quality_gate_idle_all_features	intake_connections	10/10	bounds checks dashboard
✅	quality_gate_idle_all_features	memory_usage	10/10	bounds checks dashboard
✅	quality_gate_logs	intake_connections	10/10
✅	quality_gate_logs	lost_bytes	10/10
✅	quality_gate_logs	memory_usage	10/10

Explanation

Confidence level: 90.00%
Effect size tolerance: |Δ mean %| ≥ 5.00%

Performance changes are noted in the perf column of each table:

• ✅ = significantly better comparison variant performance
• ❌ = significantly worse comparison variant performance
• ➖ = no significant change in performance

A regression test is an A/B test of target performance in a repeatable rig, where "performance" is measured as "comparison variant minus baseline variant" for an optimization goal (e.g., ingress throughput). Due to intrinsic variability in measuring that goal, we can only estimate its mean value for each experiment; we report uncertainty in that value as a 90.00% confidence interval denoted "Δ mean % CI".

For each experiment, we decide whether a change in performance is a "regression" -- a change worth investigating further -- if all of the following criteria are true:

1. Its estimated |Δ mean %| ≥ 5.00%, indicating the change is big enough to merit a closer look.

2. Its 90.00% confidence interval "Δ mean % CI" does not contain zero, indicating that if our statistical model is accurate, there is at least a 90.00% chance there is a difference in performance between baseline and comparison variants.

3. Its configuration does not mark it "erratic".

CI Pass/Fail Decision

✅ Passed. All Quality Gates passed.

• quality_gate_idle, bounds check memory_usage: 10/10 replicas passed. Gate passed.
• quality_gate_idle, bounds check intake_connections: 10/10 replicas passed. Gate passed.
• quality_gate_logs, bounds check memory_usage: 10/10 replicas passed. Gate passed.
• quality_gate_logs, bounds check intake_connections: 10/10 replicas passed. Gate passed.
• quality_gate_logs, bounds check lost_bytes: 10/10 replicas passed. Gate passed.
• quality_gate_idle_all_features, bounds check intake_connections: 10/10 replicas passed. Gate passed.
• quality_gate_idle_all_features, bounds check memory_usage: 10/10 replicas passed. Gate passed.

[AlexandreYang]

@pimlu Shall we use "golang.org/x/time/rate" to avoid reimplementing rate limiting logic, this seems to be already used in multiple places in datadog-agent.

[pimlu]

Thanks for the suggestion, golang.org/x/time/rate is awesome - I changed this to use it. It also lets us pass in mocked times which is great because the tests still work nicely.

[AlexandreYang]

1500 traceroutes per min seems to be quite high, it's 25 traceroute per sec.

For comparison, if 1 traceroute takes about 2sec, with 4 workers (default) are are about 120 traceroute per sec ((60/2)*4).

So, maybe, 150 traceroutes per host per minute might be a better default?

[pimlu]

Good callout, updated to 150!