Add Slurm-job GPU tagging (system probe)

[zhuminyi]

What does this PR do?

This PR adds Slurm job attribution to GPU process metrics.

When gpu.slurm_job_tagging.enabled is enabled, the GPU check resolves the Slurm job identity for each GPU-using process and adds the following tags when available:

slurm_job_id
slurm_job_name
slurm_job_partition

The resolution is done from the node Agent by reading SLURM_JOB_* variables from /proc/<pid>/environ. This avoids requiring a Slurm binaries, munge access, or hostPID. The only additional runtime requirement is SYS_PTRACE, which is needed for the Agent to read another process’s environment.

The feature is opt-in and disabled by default, so existing GPU check behavior is unchanged unless gpu.slurm_job_tagging.enabled is explicitly enabled.

Motivation

In Slurm-on-Kubernetes deployments such as SUNK/Slinky, Slurm jobs run as host processes inside compute pods rather than as individual Kubernetes pods. As a result, the GPU check can see the GPU-using process, but cannot attribute it to the Slurm job that owns it.

Today, solving that attribution requires a compute sidecar with Slurm tooling and access to the job PID namespace. This PR provides a simpler node-Agent-only mechanism by using the Slurm environment variables that are already present on launched job processes.

This lets GPU process metrics be correlated directly with Slurm jobs without changing the compute pod model.

Implementation details

This PR introduces a new Linux-only package, pkg/process/util/slurm, that resolves Slurm job identity for a PID from /proc/<pid>/environ.

It is wired into the existing GPU WorkloadTagCache as an additive tagging step, independent of existing container tag resolution. This is important because a Slurm job process may or may not resolve to Kubernetes container metadata.

Permission failures are treated differently from normal misses:

• no SLURM_JOB_ID means “not a Slurm job”
• permission denied means the Agent is likely missing SYS_PTRACE

Companion change

On its own, this PR only stamps slurm_job_id / slurm_job_name / slurm_job_partition onto GPU-process metrics. Those tags are only useful if there is a matching set of Slurm job/node/partition metrics carrying the same slurm_job_* tags to correlate against.

That correlation data comes from the Slurm integration. In a Slurm-on-Kubernetes deployment, the Slurm binaries live inside the Slurm pods, not on the node Agent, so the CLI check would need a agent sidecar into Slurm pods. (Note, this is not recommended, see public doc)
On the other hand, The companion integrations-core#24566 — DataDog/integrations-core#24566 adds a REST (slurmrestd) collection mode to the Slurm integration so it can be scheduled as a cluster check against the slurm-restapi service, with no sidecar and no local Slurm binaries.

How to enable

Both of the following are required for the GPU tagging to run:

• gpu.enabled (DD_GPU_ENABLED=true) — the GPU check itself
• gpu.slurm_job_tagging.enabled (DD_GPU_SLURM_JOB_TAGGING_ENABLED=true) — this feature

This is a node Agent config flag, it applies to the GPU check on every node Agent where it is set.
Helm values:

  datadog:
    env:
      - name: DD_GPU_SLURM_JOB_TAGGING_ENABLED
        value: "true"

Datadog Operator (DatadogAgent CR):

  spec:
    override:
      nodeAgent:
        env:
          - name: DD_GPU_SLURM_JOB_TAGGING_ENABLED
            value: "true" 

Describe how you validated your changes

[github-actions]

@codex review

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 7814659e88

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

[dd-octo-sts]

Go Package Import Differences

Baseline: f88736c
Comparison: 4032761

binary	os	arch	change
system-probe	linux	amd64	+1, -0 +github.com/DataDog/datadog-agent/pkg/process/util/slurm
system-probe	linux	arm64	+1, -0 +github.com/DataDog/datadog-agent/pkg/process/util/slurm

[datadog-prod-us1-6]

🎯 Code Coverage (details)
• Patch Coverage: 80.49%
• Overall Coverage: 51.84% (-0.04%)

_{This comment will be updated automatically if new data arrives.
🔗 Commit SHA: 4032761 | Docs | Datadog PR Page | Give us feedback!}

[dd-octo-sts]

Files inventory check summary

File checks results against ancestor f88736cb:

Results for datadog-agent_7.83.0~devel.git.135.4032761.pipeline.125188644-1_amd64.deb:

No change detected

[dd-octo-sts]

Static quality checks

✅ Please find below the results from static quality gates
Comparison made with ancestor f88736c
📊 Static Quality Gates Dashboard
🔗 SQG Job

Successful checks

Info

	Quality gate	Change	Size (prev → curr → max)
✅	agent_deb_amd64	+12.0 KiB (0.00% increase, -0.16% of buffer)	750.968 → 750.979 → 758.200
✅	agent_deb_amd64_fips	+16.0 KiB (0.00% increase, -0.39% of buffer)	705.876 → 705.892 → 709.840
✅	agent_rpm_amd64	+12.0 KiB (0.00% increase, -0.16% of buffer)	750.951 → 750.963 → 758.170
✅	agent_rpm_amd64_fips	+16.0 KiB (0.00% increase, -0.39% of buffer)	705.860 → 705.876 → 709.840
✅	agent_rpm_arm64	+20.0 KiB (0.00% increase, -0.54% of buffer)	726.035 → 726.055 → 729.660
✅	agent_rpm_arm64_fips	+16.0 KiB (0.00% increase, -0.37% of buffer)	684.625 → 684.641 → 688.860
✅	agent_suse_amd64	+12.0 KiB (0.00% increase, -0.16% of buffer)	750.951 → 750.963 → 758.170
✅	agent_suse_amd64_fips	+16.0 KiB (0.00% increase, -0.39% of buffer)	705.860 → 705.876 → 709.840
✅	agent_suse_arm64	+20.0 KiB (0.00% increase, -0.54% of buffer)	726.035 → 726.055 → 729.660
✅	agent_suse_arm64_fips	+16.0 KiB (0.00% increase, -0.37% of buffer)	684.625 → 684.641 → 688.860
✅	docker_agent_amd64	+12.14 KiB (0.00% increase, -0.29% of buffer)	809.674 → 809.686 → 813.790
✅	docker_agent_arm64	+20.13 KiB (0.00% increase, -0.37% of buffer)	809.763 → 809.783 → 815.030
✅	docker_agent_jmx_amd64	+12.13 KiB (0.00% increase, -0.30% of buffer)	1000.572 → 1000.584 → 1004.550
✅	docker_agent_jmx_arm64	+20.14 KiB (0.00% increase, -0.36% of buffer)	989.313 → 989.332 → 994.710
✅	docker_dogstatsd_arm64	-64.01 KiB (0.17% reduction, +6.93% of buffer)	37.367 → 37.305 → 38.270

18 successful checks with minimal change (< 2 KiB)

	Quality gate	Current Size
✅	agent_heroku_amd64	307.697 MiB
✅	agent_msi	645.641 MiB
✅	docker_cluster_agent_amd64	209.881 MiB
✅	docker_cluster_agent_arm64	222.952 MiB
✅	docker_cws_instrumentation_amd64	7.447 MiB
✅	docker_cws_instrumentation_arm64	6.877 MiB
✅	docker_dogstatsd_amd64	39.216 MiB
✅	docker_host_profiler_amd64	303.092 MiB
✅	docker_host_profiler_arm64	314.591 MiB
✅	dogstatsd_deb_amd64	29.951 MiB
✅	dogstatsd_deb_arm64	27.994 MiB
✅	dogstatsd_rpm_amd64	29.951 MiB
✅	dogstatsd_suse_amd64	29.951 MiB
✅	iot_agent_deb_amd64	46.194 MiB
✅	iot_agent_deb_arm64	42.882 MiB
✅	iot_agent_deb_armhf	43.647 MiB
✅	iot_agent_rpm_amd64	46.195 MiB
✅	iot_agent_suse_amd64	46.194 MiB

[cit-pr-commenter-54b7da]

Regression Detector

Regression Detector Results

Metrics dashboard
Target profiles
Run ID: 54068a12-d0a7-4c14-bfcf-96b34852b308

Baseline: f88736c
Comparison: 4032761
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_security_idle	memory utilization	+0.18	[+0.13, +0.24]	1	Logs bounds checks dashboard
➖	quality_gate_metrics_logs	memory utilization	+0.04	[-0.21, +0.28]	1	Logs bounds checks dashboard
➖	quality_gate_idle_all_features	memory utilization	-0.15	[-0.22, -0.08]	1	Logs bounds checks dashboard
➖	quality_gate_security_no_fs_load	memory utilization	-0.28	[-0.38, -0.19]	1	Logs bounds checks dashboard
➖	quality_gate_security_mean_fs_load	memory utilization	-0.37	[-0.40, -0.33]	1	Logs bounds checks dashboard
➖	quality_gate_logs	% cpu utilization	-0.39	[-1.35, +0.57]	1	Logs bounds checks dashboard
➖	quality_gate_idle	memory utilization	-0.43	[-0.48, -0.38]	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	147.16MiB ≤ 154MiB	bounds checks dashboard
✅	quality_gate_idle	total_bytes_received	10/10	731.32KiB ≤ 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	492.78MiB ≤ 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	181.65MiB ≤ 195MiB	bounds checks dashboard
✅	quality_gate_logs	missed_bytes	10/10	0B = 0B	bounds checks dashboard
✅	quality_gate_logs	total_bytes_received	10/10	264.14MiB ≤ 292MiB	bounds checks dashboard
✅	quality_gate_metrics_logs	cpu_usage	10/10	370.41 ≤ 2000	bounds checks dashboard
✅	quality_gate_metrics_logs	intake_connections	10/10	3 ≤ 6	bounds checks dashboard
✅	quality_gate_metrics_logs	memory_usage	10/10	406.65MiB ≤ 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
✅	quality_gate_security_idle	cpu_usage	10/10	38.17 ≤ 40	bounds checks dashboard
✅	quality_gate_security_idle	memory_usage	10/10	300.64MiB ≤ 330MiB	bounds checks dashboard
✅	quality_gate_security_mean_fs_load	cpu_usage	10/10	64.12 ≤ 80	bounds checks dashboard
✅	quality_gate_security_mean_fs_load	memory_usage	10/10	275.55MiB ≤ 310MiB	bounds checks dashboard
✅	quality_gate_security_no_fs_load	cpu_usage	10/10	23.19 ≤ 40	bounds checks dashboard
✅	quality_gate_security_no_fs_load	memory_usage	10/10	289.66MiB ≤ 320MiB	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 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	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_logs	baseline	10	Oom killed	Debug Dashboard
quality_gate_logs	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
quality_gate_security_idle	baseline	10	Oom killed	Debug Dashboard
quality_gate_security_idle	comparison	10	Crashed (exit code: 134)	Debug Dashboard
quality_gate_security_mean_fs_load	baseline	10	Crashed (exit code: 134)	Debug Dashboard
quality_gate_security_mean_fs_load	comparison	10	Oom killed	Debug Dashboard
quality_gate_security_no_fs_load	baseline	10	Oom killed	Debug Dashboard
quality_gate_security_no_fs_load	comparison	10	Crashed (exit code: 134)	Debug Dashboard

CI Pass/Fail Decision

✅ Passed. All Quality Gates passed.

• quality_gate_security_mean_fs_load, bounds check memory_usage: 10/10 replicas passed. Gate passed.
• quality_gate_security_mean_fs_load, bounds check cpu_usage: 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_security_no_fs_load, bounds check cpu_usage: 10/10 replicas passed. Gate passed.
• quality_gate_security_no_fs_load, bounds check memory_usage: 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, bounds check total_bytes_received: 10/10 replicas passed. Gate passed.
• quality_gate_security_idle, bounds check memory_usage: 10/10 replicas passed. Gate passed.
• quality_gate_security_idle, 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 cpu_usage: 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 intake_connections: 10/10 replicas passed. Gate passed.
• quality_gate_idle_all_features, bounds check total_bytes_received: 10/10 replicas passed. Gate passed.
• quality_gate_idle_all_features, bounds check memory_usage: 10/10 replicas passed. Gate passed.