Autodiscovery Config Filtering via CEL

[gabedos]

What does this PR do?

Introduces a new matching capability for check-configurations and services within Autodiscovery. The cel_selector takes in rules defined by Google's Common Expression Language and allows more granular decision making onto which service a check gets scheduled on.

The cel_selector will be able to be configured independently and also with traditional ad_identifiers / advanced_ad_identifiers:

1. If used alone, the cel_selector will be used to compare every candidate service with the check-config (more computational expensive and can be risky).
2. If the cel_selector is defined with an ADID, then the Autodiscovery Identifier will be applied first. Only if the ADIDs match between the service and config will the cel_selector execute to perform more granular matching.

The support of the cel_selector in Autodiscovery is limited to containers, kube_services, kube_endpoints, and in the near future processes. The structures for each entity are defined here in the WorkloadFilter proto definitions. This still remains to be documented.

Motivation

Expose more granular check scheduling capabilities for the Agent. CONTP-984

Describe how you validated your changes

Deploy the Agent with the subsequent helm chart.

• On the Node Agent, it contains a check config which targets all containers in the datadog-agent namespace. It also contains a check config which targets an nginx image in a container called load-balancer.
• On the Cluster Agent, it container a check config which targets my-nginx service in the default namespace with datadog-obs annotation. It also contains a check config which targets all services in the cluster.

datadog:
  kubelet:
    tlsVerify: false
  clusterChecks:
    enabled: true
  confd:
    # Schedules redisdb check on every container in the "datadog-agent" namespace
    redisdb.yaml: |-
      cel_selector:
        containers:
          - container.pod.namespace == "datadog-agent" && container.image.reference != ""
      init_config:
      instances:
        - host: "%%host%%"
          port: "%%port%%"
          name: "gabedos-test-cel"
    # Schedules http_check on load-balancer container running nginx image
    http_check.yaml: |-
      ad_identifiers:
        - nginx
      cel_selector:
        containers:
          - container.name.matches("load-balancer")
      init_config:
      instances:
        - url: "http://%%host%%"
          name: "my-cel-nginx"
clusterAgent:
  enabled: true
  replicas: 1
  confd:
    http_check.yaml: |-
      advanced_ad_identifiers:
        - kube_service:
            name: "my-nginx"
            namespace: "default"
      cel_selector:
        kube_services:
          - kube_service.annotations["datadog-obs"] == "true"
      cluster_check: true
      init_config:
      instances:
        - url: "http://%%host%%"
          name: "my-nginx"
    http_check2.yaml: |-
      cel_selector:
        kube_services:
          - kube_service.name.matches("") || kube_service.namespace != ""
      cluster_check: true
      init_config:
      instances:
        - url: "http://%%host%%"
          name: "all-services-check"

Case 1 (Node + CEL + File): On the node agent, you should expect agent configcheck to expose a redisdb check configured for all containers in the datadog-agent namespace (node agent, dca, and clc runners).

Case 2 (Node + ADID + CEL + File): On the node agent, you should expect `agent configcheck` to expose a http check configured for the nginx pod.

apiVersion: v1
kind: Pod
metadata:
  name: nginx-pod
spec:
  containers:
    - name: load-balancer
      image: nginx:latest
      ports:
        - containerPort: 80

Case 3 (DCA + CEL + File): On the cluster agent, you should expect agent configcheck to expose a http check2 configured for every single service in your cluster.

Case 4 (DCA + ADID + CEL + File): On the cluster agent, you should expect `agent configcheck` to expose a http check configured for the service defined below.

apiVersion: v1
kind: Service
metadata:
  name: my-nginx
  namespace: default
  annotations:
    team: "platform-team"
    datadog-obs: "true"
spec:
  selector:
    app: nginx
  ports:
    - protocol: TCP
      port: 80
      targetPort: 80
  type: ClusterIP

Additional Notes

CELSelector via Annotations: We have decided to not implement support for parsing this rule when defining check configuration via annotations/labels on a resource. We assume that if you are explicitly adding a check config directly onto a granular resource, the user would be wanting to run the check. This simplifies Agent side code as well as reduces risks of misconfiguration from users.

DCA KubeServices + KubeEndpoints: We want to allow users to filter on services and endpoints based on their parameters like name, namespace, and annotations. Currently in the cluster-agent we do not store a complete view of these entities. Autodiscovery has its own direct collector from the APIServer. Therefore, we must store the metadata of the k8s object within the Autodiscovery service entity. This is limited to the name, namespace, and annotations of the entity.

Performance: This change was tested and profiled on an experimental cluster. We can see that in terms of CPU usage there are no drastic changes. The service reconciliation only happens when new workloads enter the node which should be relatively infrequent. In terms of memory usage, the compiled regex expression accounts for about half the memory usage for the compiled program and the other half is other miscellaneous costs of CEL. Memory costs linearly scale with the size of rules which users define.

CPU Profiles

Memory Profiles

TODO Remaining Work:

1. Re-evaluate whether we want to restrict users to define rules that have granular identifiers. For container checks this means including a short image name. For services/endpoints this means including a name and namespace in the conditions.

[cit-pr-commenter]

Regression Detector

Regression Detector Results

Metrics dashboard
Target profiles
Run ID: d813a414-7c7f-45a9-b997-cd5000596e40

Baseline: 8497514
Comparison: f8a0d60
Diff

Optimization Goals: ✅ No significant changes detected

Experiments ignored for regressions

Regressions in experiments with settings containing erratic: true are ignored.

perf	experiment	goal	Δ mean %	Δ mean % CI	trials	links
✅	docker_containers_cpu	% cpu utilization	-23.81	[-25.43, -22.18]	1	Logs

Fine details of change detection per experiment

perf	experiment	goal	Δ mean %	Δ mean % CI	trials	links
➖	quality_gate_metrics_logs	memory utilization	+1.11	[+0.89, +1.32]	1	Logs bounds checks dashboard
➖	tcp_syslog_to_blackhole	ingress throughput	+0.70	[+0.62, +0.79]	1	Logs
➖	otlp_ingest_metrics	memory utilization	+0.39	[+0.26, +0.51]	1	Logs
➖	file_tree	memory utilization	+0.37	[+0.33, +0.41]	1	Logs
➖	quality_gate_idle	memory utilization	+0.18	[+0.13, +0.24]	1	Logs bounds checks dashboard
➖	ddot_metrics_sum_delta	memory utilization	+0.07	[-0.09, +0.22]	1	Logs
➖	ddot_logs	memory utilization	+0.07	[-0.00, +0.14]	1	Logs
➖	file_to_blackhole_100ms_latency	egress throughput	+0.06	[-0.54, +0.67]	1	Logs
➖	file_to_blackhole_0ms_latency	egress throughput	+0.05	[-0.55, +0.66]	1	Logs
➖	file_to_blackhole_500ms_latency	egress throughput	+0.02	[-0.59, +0.64]	1	Logs
➖	uds_dogstatsd_to_api	ingress throughput	+0.00	[-0.21, +0.21]	1	Logs
➖	tcp_dd_logs_filter_exclude	ingress throughput	+0.00	[-0.01, +0.01]	1	Logs
➖	uds_dogstatsd_20mb_12k_contexts_20_senders	memory utilization	-0.01	[-0.06, +0.03]	1	Logs
➖	quality_gate_idle_all_features	memory utilization	-0.03	[-0.07, +0.01]	1	Logs bounds checks dashboard
➖	file_to_blackhole_1000ms_latency	egress throughput	-0.12	[-0.74, +0.50]	1	Logs
➖	otlp_ingest_logs	memory utilization	-0.15	[-0.29, -0.02]	1	Logs
➖	ddot_metrics_sum_cumulativetodelta_exporter	memory utilization	-0.24	[-0.43, -0.05]	1	Logs
➖	docker_containers_memory	memory utilization	-0.30	[-0.74, +0.15]	1	Logs
➖	ddot_metrics	memory utilization	-0.57	[-0.73, -0.40]	1	Logs
➖	ddot_metrics_sum_cumulative	memory utilization	-0.62	[-0.73, -0.51]	1	Logs
➖	quality_gate_logs	% cpu utilization	-2.31	[-5.03, +0.42]	1	Logs bounds checks dashboard
✅	docker_containers_cpu	% cpu utilization	-23.81	[-25.43, -22.18]	1	Logs

Bounds Checks: ❌ Failed

perf	experiment	bounds_check_name	replicates_passed	links
✅	docker_containers_cpu	simple_check_run	10/10
❌	docker_containers_memory	memory_usage	9/10
✅	docker_containers_memory	simple_check_run	10/10
✅	file_to_blackhole_0ms_latency	lost_bytes	10/10
✅	file_to_blackhole_0ms_latency	memory_usage	10/10
✅	file_to_blackhole_1000ms_latency	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_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	bounds checks dashboard
✅	quality_gate_logs	lost_bytes	10/10	bounds checks dashboard
✅	quality_gate_logs	memory_usage	10/10	bounds checks dashboard
✅	quality_gate_metrics_logs	cpu_usage	10/10	bounds checks dashboard
✅	quality_gate_metrics_logs	intake_connections	10/10	bounds checks dashboard
✅	quality_gate_metrics_logs	lost_bytes	10/10	bounds checks dashboard
✅	quality_gate_metrics_logs	memory_usage	10/10	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".

CI Pass/Fail Decision

✅ Passed. All Quality Gates passed.

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

[agent-platform-auto-pr]

Static quality checks

✅ Please find below the results from static quality gates
Comparison made with ancestor 8497514

Successful checks

Info

	Quality gate	Delta	On disk size (MiB)	Delta	On wire size (MiB)
✅	agent_deb_amd64	$${+0.02}$$	$${676.29}$$ < $${702.15}$$	$${-0.01}$$	$${165.14}$$ < $${177.79}$$
✅	agent_deb_amd64_fips	$${+0.02}$$	$${670.8}$$ < $${696.65}$$	$${-0.04}$$	$${164.59}$$ < $${176.53}$$
✅	agent_heroku_amd64	$${+0.01}$$	$${336.63}$$ < $${340.18}$$	$${-0.01}$$	$${89.84}$$ < $${91.08}$$
✅	agent_msi	$${+0.06}$$	$${1013.83}$$ < $${1015.38}$$	$${-0.01}$$	$${148.42}$$ < $${150.78}$$
✅	agent_rpm_amd64	$${+0.02}$$	$${676.28}$$ < $${702.14}$$	$${-0.04}$$	$${167.39}$$ < $${180.53}$$
✅	agent_rpm_amd64_fips	$${+0.02}$$	$${670.79}$$ < $${696.64}$$	$${+0.02}$$	$${166.05}$$ < $${178.79}$$
✅	agent_rpm_arm64	$${+0.02}$$	$${666.09}$$ < $${686.31}$$	$${-0.01}$$	$${153.33}$$ < $${161.22}$$
✅	agent_rpm_arm64_fips	$${+0.02}$$	$${661.61}$$ < $${681.84}$$	$${+0.02}$$	$${152.68}$$ < $${160.18}$$
✅	agent_suse_amd64	$${+0.02}$$	$${676.28}$$ < $${702.14}$$	$${-0.04}$$	$${167.39}$$ < $${180.53}$$
✅	agent_suse_amd64_fips	$${+0.02}$$	$${670.79}$$ < $${696.64}$$	$${+0.02}$$	$${166.05}$$ < $${178.79}$$
✅	agent_suse_arm64	$${+0.02}$$	$${666.09}$$ < $${686.31}$$	$${-0.01}$$	$${153.33}$$ < $${161.22}$$
✅	agent_suse_arm64_fips	$${+0.02}$$	$${661.61}$$ < $${681.84}$$	$${+0.02}$$	$${152.68}$$ < $${160.18}$$
✅	docker_agent_amd64	$${+0.02}$$	$${746.84}$$ < $${773.59}$$	$${-0}$$	$${251.85}$$ < $${266.06}$$
✅	docker_agent_arm64	$${+0.02}$$	$${760.6}$$ < $${781.7}$$	$${+0.02}$$	$${242.46}$$ < $${251.85}$$
✅	docker_agent_jmx_amd64	$${+0.02}$$	$${937.71}$$ < $${964.45}$$	$${+0.03}$$	$${320.49}$$ < $${334.68}$$
✅	docker_agent_jmx_arm64	$${+0.02}$$	$${940.06}$$ < $${961.17}$$	$${-0.02}$$	$${307.04}$$ < $${316.44}$$
✅	docker_cluster_agent_amd64	$${+0.02}$$	$${213.03}$$ < $${213.74}$$	$${+0}$$	$${72.27}$$ < $${73.14}$$
✅	docker_cluster_agent_arm64	$${0}$$	$${228.92}$$ < $${229.68}$$	$${+0.01}$$	$${68.53}$$ < $${69.41}$$
✅	docker_cws_instrumentation_amd64	$${0}$$	$${7.07}$$ < $${7.12}$$	$${-0}$$	$${2.95}$$ < $${3.29}$$
✅	docker_cws_instrumentation_arm64	$${-0}$$	$${6.69}$$ < $${6.92}$$	$${-0}$$	$${2.7}$$ < $${3.07}$$
✅	docker_dogstatsd_amd64	$${+0}$$	$${38.45}$$ < $${39.3}$$	$${+0}$$	$${14.84}$$ < $${15.76}$$
✅	docker_dogstatsd_arm64	$${-0}$$	$${37.12}$$ < $${37.94}$$	$${-0.01}$$	$${14.28}$$ < $${14.83}$$
✅	dogstatsd_deb_amd64	$${+0}$$	$${29.67}$$ < $${30.53}$$	$${+0}$$	$${7.82}$$ < $${8.75}$$
✅	dogstatsd_deb_arm64	$${+0}$$	$${28.25}$$ < $${29.11}$$	$${-0}$$	$${6.77}$$ < $${7.71}$$
✅	dogstatsd_rpm_amd64	$${+0}$$	$${29.67}$$ < $${30.53}$$	$${+0}$$	$${7.83}$$ < $${8.76}$$
✅	dogstatsd_suse_amd64	$${+0}$$	$${29.67}$$ < $${30.53}$$	$${+0}$$	$${7.83}$$ < $${8.76}$$
✅	iot_agent_deb_amd64	$${+0.01}$$	$${41.92}$$ < $${54.97}$$	$${+0}$$	$${10.9}$$ < $${14.45}$$
✅	iot_agent_deb_arm64	$${+0.01}$$	$${39.74}$$ < $${51.9}$$	$${+0}$$	$${9.42}$$ < $${12.63}$$
✅	iot_agent_deb_armhf	$${+0.01}$$	$${39.61}$$ < $${51.84}$$	$${+0}$$	$${9.5}$$ < $${12.74}$$
✅	iot_agent_rpm_amd64	$${+0.01}$$	$${41.92}$$ < $${54.97}$$	$${+0}$$	$${10.92}$$ < $${14.47}$$
✅	iot_agent_suse_amd64	$${+0.01}$$	$${41.92}$$ < $${54.97}$$	$${+0}$$	$${10.92}$$ < $${14.47}$$

[aqian01]

LGTM for files owned by ALP

[adel121]

Left a small comment

[gabedos]

/merge

[dd-devflow-routing-codex]

View all feedbacks in Devflow UI.

2025-10-21 17:44:00 UTC ℹ️ Start processing command /merge

---

2025-10-21 17:44:07 UTC ℹ️ MergeQueue: pull request added to the queue

The expected merge time in main is approximately 41m (p90).

---

2025-10-21 18:09:52 UTC ℹ️ MergeQueue: This merge request was merged