Improve kept/seen metrics for trace sampler

[keisku]

What does this PR do?

• Add tags such as sample_service and sample_env.
• The timing of the following metrics have been changed.
  • datadog.trace_agent.sampler.kept
  • datadog.trace_agent.sampler.seen

Motivation

To get more insights during troubleshooting of sampler behavior.

The timing of the following metrics have been changed.

Before this PR, when short-circuit(L112-L120) gets hit, datadog.trace_agent.sampler.kept and datadog.trace_agent.sampler.seen were not counted.
Even in this case, trace agent "sees" a trace and determins whether to "keep" a trace based on sampled from GetSamplingPriority(trace).

datadog-agent/pkg/trace/sampler/prioritysampler.go

Lines 99 to 132 in cb0d65b

	// Sample counts an incoming trace and returns the trace sampling decision and the applied sampling rate
	func (s *PrioritySampler) Sample(now time.Time, trace *pb.TraceChunk, root *pb.Span, tracerEnv string, clientDroppedP0sWeight float64) bool {
	// Extra safety, just in case one trace is empty
	if len(trace.Spans) == 0 {
	return false
	}
	samplingPriority, _ := GetSamplingPriority(trace)
	// Regardless of rates, sampling here is based on the metadata set
	// by the client library. Which, is turn, is based on agent hints,
	// but the rule of thumb is: respect client choice.
	sampled := samplingPriority > 0
	// Short-circuit and return without counting the trace in the sampling rate logic
	// if its value has not been set automatically by the client lib.
	// The feedback loop should be scoped to the values it can act upon.
	if samplingPriority < 0 {
	return sampled
	}
	if samplingPriority > 1 {
	return sampled
	}
	signature := s.catalog.register(ServiceSignature{Name: root.Service, Env: toSamplerEnv(tracerEnv, s.agentEnv)})
	// Update sampler state by counting this trace
	s.countSignature(now, root, signature, clientDroppedP0sWeight)
	if sampled {
	s.applyRate(root, signature)
	s.sampler.countSample()
	}
	return sampled
	}

Describe how you validated your changes

Using datadog/agent-dev:keisku-sampler-metrics-py3.

Priority / Error Sampler

Now we can confirm DD_TRACE_SAMPLING_RULES=[{"resource":"*","service":"note-client","sample_rate":1}] is applied with this metric.

Applied the config to tracers through Manage Ingestion Rate.

Probabilistic Sampler

Update DD_APM_PROBABILISTIC_SAMPLER_SAMPLING_PERCENTAGE from 100 to 50.

Possible Drawbacks / Trade-offs

Additional Notes

[agent-platform-auto-pr]

Uncompressed package size comparison

Comparison with ancestor 6c020c8323d53d725bef9f07f0f5a9d06111ffbe

Diff per package

package	diff	status	size	ancestor	threshold
datadog-dogstatsd-amd64-deb	0.00MB	✅	58.93MB	58.93MB	0.50MB
datadog-dogstatsd-x86_64-rpm	0.00MB	✅	59.01MB	59.01MB	0.50MB
datadog-dogstatsd-x86_64-suse	0.00MB	✅	59.01MB	59.01MB	0.50MB
datadog-dogstatsd-arm64-deb	0.00MB	✅	56.44MB	56.44MB	0.50MB
datadog-iot-agent-amd64-deb	0.00MB	✅	93.99MB	93.99MB	0.50MB
datadog-iot-agent-x86_64-rpm	0.00MB	✅	94.06MB	94.06MB	0.50MB
datadog-iot-agent-x86_64-suse	0.00MB	✅	94.06MB	94.06MB	0.50MB
datadog-iot-agent-arm64-deb	0.00MB	✅	90.04MB	90.04MB	0.50MB
datadog-iot-agent-aarch64-rpm	0.00MB	✅	90.11MB	90.11MB	0.50MB
datadog-agent-amd64-deb	-0.00MB	✅	929.12MB	929.12MB	0.50MB
datadog-agent-x86_64-rpm	-0.00MB	✅	938.78MB	938.78MB	0.50MB
datadog-agent-x86_64-suse	-0.00MB	✅	938.78MB	938.78MB	0.50MB
datadog-agent-arm64-deb	-0.00MB	✅	915.76MB	915.76MB	0.50MB
datadog-agent-aarch64-rpm	-0.00MB	✅	925.40MB	925.41MB	0.50MB
datadog-heroku-agent-amd64-deb	-0.00MB	✅	477.42MB	477.43MB	0.50MB

Decision

✅ Passed

[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=53518388 --os-family=ubuntu

Note: This applies to commit 96d9599

[cit-pr-commenter]

Regression Detector

Regression Detector Results

Metrics dashboard
Target profiles
Run ID: 66c626bb-2d95-4d3e-ab09-54cd56883a2f

Baseline: 6c020c8
Comparison: 96d9599
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	+3.03	[-0.22, +6.28]	1	Logs
➖	file_tree	memory utilization	+0.71	[+0.55, +0.87]	1	Logs
➖	file_to_blackhole_1000ms_latency_linear_load	egress throughput	+0.09	[-0.38, +0.57]	1	Logs
➖	uds_dogstatsd_to_api	ingress throughput	+0.01	[-0.09, +0.11]	1	Logs
➖	tcp_dd_logs_filter_exclude	ingress throughput	-0.00	[-0.01, +0.01]	1	Logs
➖	file_to_blackhole_100ms_latency	egress throughput	-0.01	[-0.79, +0.77]	1	Logs
➖	file_to_blackhole_0ms_latency	egress throughput	-0.02	[-0.92, +0.88]	1	Logs
➖	file_to_blackhole_300ms_latency	egress throughput	-0.04	[-0.68, +0.59]	1	Logs
➖	file_to_blackhole_0ms_latency_http1	egress throughput	-0.05	[-0.94, +0.85]	1	Logs
➖	file_to_blackhole_0ms_latency_http2	egress throughput	-0.05	[-0.91, +0.81]	1	Logs
➖	file_to_blackhole_500ms_latency	egress throughput	-0.08	[-0.88, +0.71]	1	Logs
➖	uds_dogstatsd_to_api_cpu	% cpu utilization	-0.12	[-0.83, +0.59]	1	Logs
➖	file_to_blackhole_1000ms_latency	egress throughput	-0.30	[-1.12, +0.51]	1	Logs
➖	quality_gate_idle	memory utilization	-0.37	[-0.41, -0.33]	1	Logs bounds checks dashboard
➖	quality_gate_idle_all_features	memory utilization	-0.62	[-0.71, -0.52]	1	Logs bounds checks dashboard
➖	tcp_syslog_to_blackhole	ingress throughput	-0.78	[-0.84, -0.71]	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_logs, 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 lost_bytes: 10/10 replicas passed. Gate 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_idle_all_features, bounds check memory_usage: 10/10 replicas passed. Gate passed.
• quality_gate_idle_all_features, bounds check intake_connections: 10/10 replicas passed. Gate passed.

[keisku]

We need to keep metrics aggregation on the client side(trace-agent).

datadog-agent/comp/dogstatsd/statsd/statsd.go

Lines 87 to 88 in 763d6e8

	ddgostatsd.WithClientSideAggregation(),
	ddgostatsd.WithExtendedClientSideAggregation(),

I go with #33214