[PROF-15021][Host Profiler] Deployment guides

[theomagellan]

What does this PR do?

Adds deployment documentation and manifests for the host profiler preview under cmd/host-profiler/deploy/. The docs cover all four supported deployment paths:

• Bundled mode:

  • Helm (tested on datadog-3.220.0): add hostProfiler values to existing Datadog Helm deployments. The chart configures capabilities and seccomp automatically.
  • Operator (tested on v.0.127.0): annotate existing DatadogAgent CR. Capabilities must be applied manually via a patch; seccomp requires manual node provisioning.

• Standalone mode:

  • Helm: deploy via the open-telemetry/opentelemetry-collector chart
  • Operator: deploy via an OpenTelemetryCollector CR managed by the OTel Operator

• Both Standalone paths:

  • provide value overrides to remove privileges and add capabilities through securityContext and add an init container to install seccomp profile
  • apply a standard NetworkPolicy by default. An optional CiliumNetworkPolicy is provided for Cilium clusters, scoping egress to Datadog FQDNs, the kubelet API, and the Kubernetes API server.

Motivation

Introduces a deployment guide for the Host Profiler going over each deployment path with every possible configuration depending on path and flavor.

Describe how you validated your changes

All four paths were deployed end-to-end on an EKS cluster running Cilium, following the docs exactly as a user would:

• Check profiles on Datadog
• Symbol uploads were confirmed with fresh Go binaries on each path's deployment
• Standard NetworkPolicy and CiliumNetworkPolicy were both tested for both standalone paths; collector pods were restarted after each policy change to verify connectivity at startup
• Seccomp and capabilities were verified for both standalone paths (init container copies profile from image; no manual provisioning required) and for the bundled Operator path (manual provisioning documented)

Each path of each flavor was tested on:

• ubuntu 22.04
• ubuntu 24.04
• ubuntu 26.04

Additional Notes

• The bundled Operator path requires manual seccomp provisioning (no init container mechanism available) for now as v.0.127.0 doesn't contain our security efforts. They should be included in v.0.128.0-rc2 but that version still has to be released.
• I was using Theomagellan/preview docs #51880 to iterate on the docs with Codex and Claude, feel free to maybe scan through Codex comments. The large majority of them were addressed at each iteration.

[datadog-datadog-prod-us1]

✨ Fix all issues with BitsAI

⚠️ Warnings

🚦 1 Pipeline job failed

DataDog/datadog-agent | single_machine_performance-regression_detector-merge_base_check     

Useful? React with 👍 / 👎

_{This comment will be updated automatically if new data arrives.
🔗 Commit SHA: 03d09aa | Docs | Datadog PR Page | Give us feedback!}

[theomagellan]

This doc was written with v0.127.0 in mind.
If every security feature we contributed is included in the next release candidate of v.0.128.0, what do we think of recommending it before its official release?

[dd-octo-sts]

Files inventory check summary

File checks results against ancestor da0e2341:

Results for datadog-agent_7.81.0~devel.git.551.d825270.pipeline.117821134-1_amd64.deb:

No change detected

[dd-octo-sts]

Static quality checks

✅ Please find below the results from static quality gates
Comparison made with ancestor da0e234
📊 Static Quality Gates Dashboard
🔗 SQG Job

Successful checks

Info

	Quality gate	Change	Size (prev → curr → max)
✅	docker_agent_amd64	+3.12 KiB (0.00% increase, -0.30% of buffer)	812.788 → 812.791 → 813.790
✅	docker_agent_jmx_amd64	+3.13 KiB (0.00% increase, -0.37% of buffer)	1003.729 → 1003.732 → 1004.550
✅	docker_host_profiler_amd64	+6.61 KiB (0.00% increase, -0.05% of buffer)	304.503 → 304.509 → 317.640

29 successful checks with minimal change (< 2 KiB)

	Quality gate	Current Size
✅	agent_deb_amd64	753.354 MiB
✅	agent_deb_amd64_fips	709.129 MiB
✅	agent_heroku_amd64	312.146 MiB
✅	agent_rpm_amd64	753.338 MiB
✅	agent_rpm_amd64_fips	709.113 MiB
✅	agent_rpm_arm64	728.647 MiB
✅	agent_rpm_arm64_fips	688.161 MiB
✅	agent_suse_amd64	753.338 MiB
✅	agent_suse_amd64_fips	709.113 MiB
✅	agent_suse_arm64	728.647 MiB
✅	agent_suse_arm64_fips	688.161 MiB
✅	docker_agent_arm64	813.120 MiB
✅	docker_agent_jmx_arm64	992.714 MiB
✅	docker_cluster_agent_amd64	209.738 MiB
✅	docker_cluster_agent_arm64	222.847 MiB
✅	docker_cws_instrumentation_amd64	7.447 MiB
✅	docker_cws_instrumentation_arm64	6.877 MiB
✅	docker_dogstatsd_amd64	39.833 MiB
✅	docker_dogstatsd_arm64	37.884 MiB
✅	docker_host_profiler_arm64	315.634 MiB
✅	dogstatsd_deb_amd64	30.490 MiB
✅	dogstatsd_deb_arm64	28.487 MiB
✅	dogstatsd_rpm_amd64	30.490 MiB
✅	dogstatsd_suse_amd64	30.490 MiB
✅	iot_agent_deb_amd64	45.619 MiB
✅	iot_agent_deb_arm64	42.338 MiB
✅	iot_agent_deb_armhf	43.133 MiB
✅	iot_agent_rpm_amd64	45.620 MiB
✅	iot_agent_suse_amd64	45.619 MiB

[cit-pr-commenter-54b7da]

Regression Detector

Regression Detector Results

Metrics dashboard
Target profiles
Run ID: e66b0c9d-31bd-4970-b669-6dd964dcdac1

Baseline: da0e234
Comparison: d825270
Diff

❌ Experiments with retried target crashes

This is a critical error. One or more replicates failed with a non-zero exit code. These replicates may have been retried. See Replicate Execution Details for more information.

• quality_gate_idle

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_idle	memory utilization	+0.04	[-0.02, +0.10]	1	Logs bounds checks dashboard
➖	quality_gate_idle_all_features	memory utilization	-0.01	[-0.05, +0.03]	1	Logs bounds checks dashboard
➖	quality_gate_metrics_logs	memory utilization	-0.11	[-0.37, +0.14]	1	Logs bounds checks dashboard
➖	quality_gate_logs	% cpu utilization	-0.71	[-1.75, +0.32]	1	Logs bounds checks dashboard

Bounds Checks: ✅ Passed

perf	experiment	bounds_check_name	replicates_passed	observed_value	links
✅	quality_gate_idle	intake_connections	10/10	3 ≤ 4	bounds checks dashboard
✅	quality_gate_idle	memory_usage	10/10	144.46MiB ≤ 147MiB	bounds checks dashboard
✅	quality_gate_idle	total_bytes_received	10/10	730.24KiB ≤ 819.20KiB	bounds checks dashboard
✅	quality_gate_idle_all_features	intake_connections	10/10	3 ≤ 4	bounds checks dashboard
✅	quality_gate_idle_all_features	memory_usage	10/10	482.24MiB ≤ 495MiB	bounds checks dashboard
✅	quality_gate_idle_all_features	total_bytes_received	10/10	1.12MiB ≤ 1.25MiB	bounds checks dashboard
✅	quality_gate_logs	intake_connections	10/10	4 ≤ 6	bounds checks dashboard
✅	quality_gate_logs	memory_usage	10/10	180.82MiB ≤ 195MiB	bounds checks dashboard
✅	quality_gate_logs	missed_bytes	10/10	0B = 0B	bounds checks dashboard
✅	quality_gate_logs	total_bytes_received	10/10	264.03MiB ≤ 292MiB	bounds checks dashboard
✅	quality_gate_metrics_logs	cpu_usage	10/10	338.04 ≤ 2000	bounds checks dashboard
✅	quality_gate_metrics_logs	intake_connections	10/10	4 ≤ 6	bounds checks dashboard
✅	quality_gate_metrics_logs	memory_usage	10/10	392.42MiB ≤ 430MiB	bounds checks dashboard
✅	quality_gate_metrics_logs	missed_bytes	10/10	0B = 0B	bounds checks dashboard
✅	quality_gate_metrics_logs	total_bytes_received	10/10	0.93GiB ≤ 1.04GiB	bounds checks dashboard

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".

Replicate Execution Details

We run multiple replicates for each experiment/variant. However, we allow replicates to be automatically retried if there are any failures, up to 8 times, at which point the replicate is marked dead and we are unable to run analysis for the entire experiment. We call each of these attempts at running replicates a replicate execution. This section lists all replicate executions that failed due to the target crashing or being oom killed.

Note: In the below tables we bucket failures by experiment, variant, and failure type. For each of these buckets we list out the replicate indexes that failed with an annotation signifying how many times said replicate failed with the given failure mode. In the below example the baseline variant of the experiment named experiment_with_failures had two replicates that failed by oom kills. Replicate 0, which failed 8 executions, and replicate 1 which failed 6 executions, all with the same failure mode.

Experiment	Variant	Replicates	Failure	Logs	Debug Dashboard
experiment_with_failures	baseline	0 (x8) 1 (x6)	Oom killed		Debug Dashboard

The debug dashboard links will take you to a debugging dashboard specifically designed to investigate replicate execution failures.

❌ Retried Normal Replicate Execution Failures (non-profiling)

Experiment	Variant	Replicates	Failure	Debug Dashboard
quality_gate_idle	comparison	0	Oom killed	Debug Dashboard

❌ Retried Profiling Replicate Execution Failures (ddprof)

Note: Profiling replicas may still be executing. See the debug dashboard for up to date status.

Experiment	Variant	Replicates	Failure	Debug Dashboard
quality_gate_idle	baseline	10	Oom killed	Debug Dashboard
quality_gate_idle	comparison	10	Oom killed	Debug Dashboard
quality_gate_idle_all_features	baseline	10	Oom killed	Debug Dashboard
quality_gate_idle_all_features	comparison	10	Oom killed	Debug Dashboard
quality_gate_metrics_logs	baseline	10	Oom killed	Debug Dashboard
quality_gate_metrics_logs	comparison	10	Oom killed	Debug Dashboard

CI Pass/Fail Decision

✅ Passed. All Quality Gates passed.

• quality_gate_idle, bounds check total_bytes_received: 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_logs, bounds check total_bytes_received: 10/10 replicas passed. Gate passed.
• quality_gate_logs, bounds check missed_bytes: 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_metrics_logs, bounds check cpu_usage: 10/10 replicas passed. Gate passed.
• quality_gate_metrics_logs, bounds check intake_connections: 10/10 replicas passed. Gate passed.
• quality_gate_metrics_logs, bounds check missed_bytes: 10/10 replicas passed. Gate passed.
• quality_gate_metrics_logs, bounds check total_bytes_received: 10/10 replicas passed. Gate passed.
• quality_gate_metrics_logs, bounds check memory_usage: 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.
• quality_gate_idle_all_features, bounds check total_bytes_received: 10/10 replicas passed. Gate passed.

[RiantZ]

pipelines reference processor k8sattributes but it's not defined in the processors: section of this file. Works only because pod-spec.yaml enables presets.kubernetesAttributes and the Helm chart injects the config — but the processor name injected by the preset is k8sattributes, while operator/collector.yaml defines it as k8s_attributes. Inconsistent naming between the two standalone paths. Verify both resolve at runtime; if only one name is valid, the other path breaks silently.

[theomagellan]

Yes, that's a quirk I wasn't really able to come around to be able to include the same k8sattributes in both metrics and profiles pipeline. This is also relying on open-telemetry/opentelemetry-helm-charts#2242 for rewriting k8sattributes into k8s_attributes