[CONTP-48] Cluster Agent consistent tagging via global tags

[gabedos]

What does this PR do?

This PR unifies the way that data is tagged from cluster checks. All data from a cluster check (including those that are technically not cluster checks but still executed on the DCA on the cluster level such as KSM check) will receive the global tags defined on the cluster agent (DD_TAGS, DD_EXTRA_TAGS, DD_CLUSTER_CHECKS_EXTRA_TAGS, DD_ORCHESTRATOR_EXPLORER_EXTRA_TAGS).

Motivation

Simplify the way users interact with cluster check tags. Currently we have 4 different environment variables that apply differently for some checks (KSM, Orchestrator, others) and also differently depending on where the check is executed (DCA, Node Agent, CLC runner).

Describe how to test/QA your changes

Case 1: DCA w/o CLC

1. Deploy the Agent

datadog:
  kubelet:
    tlsVerify: false
  clusterName: <INSERT_CLUSTER_NAME>
  clusterChecks:
    enabled: true
  kubeStateMetricsCore:
    enabled: true
    # useClusterCheckRunners: true
  envDict:
    DD_TAGS: "source:node-tags"
    DD_EXTRA_TAGS: "source:node-extra"
    DD_CLUSTER_CHECKS_EXTRA_TAGS: "source:node-cluster_extra"       # Note: should never see these
    DD_ORCHESTRATOR_EXPLORER_EXTRA_TAGS: "source:node-orch_extra"   # Note: should never see these
agents:
  image:
    tag: <INSERT_RC_TAG>
# clusterChecksRunner:
#   replicas: 1
#   enabled: true
#   image:
#     tag: <INSERT_RC_TAG>
clusterAgent:
  envDict:
    DD_TAGS: "source:dca-tags"
    DD_EXTRA_TAGS: "source:dca-extra"
    DD_CLUSTER_CHECKS_EXTRA_TAGS: "source:dca-cluster_extra"
    DD_ORCHESTRATOR_EXPLORER_EXTRA_TAGS: "source:dca-orch_extra"
  image:
    tag: <INSERT_RC_TAG>
  enabled: true
  replicas: 1

2. Deploy the nginx service with an autodiscovery http check

apiVersion: v1
kind: Service
metadata:
    name: my-nginx
    labels:
        run: my-nginx
        tags.datadoghq.com/env: "prod"
        tags.datadoghq.com/service: "my-nginx"
        tags.datadoghq.com/version: "1.19.0"
    annotations:
      ad.datadoghq.com/service.checks: |
        {
          "http_check": {
            "init_config": {},
            "instances": [
              {
                "url":"http://%%host%%",
                "name":"My Nginx",
                "timeout":1
              }
            ]
          }
        }        
spec:
    ports:
        - port: 80
          protocol: TCP
    selector:
        run: my-nginx

3. Verify Global Tags are set

k exec -it datadog-agent-linux-cluster-agent-xxxxx -- agent tagger-list
=== Entity internal://global-entity-id ===
== Source workloadmeta-static =
=Tags: [kube_cluster_name:gabedos-cluster source:dca-cluster_extra source:dca-extra source:dca-orch_extra source:dca-tags]
===

k exec -it datadog-agent-linux-xxxxx -- agent tagger-list
=== Entity internal://global-entity-id ===
== Source workloadmeta-static =
=Tags: [source:node-extra source:node-tags]
===

4. Confirm the following metrics tags:

• network.http.cant_connect should contain all of the cluster agent tags and node agent tags because it was dispatched from the cluster agent where the DCA global tags are added to the check config and the node agent's own global tags and also enriched to the metric in the aggregator.
• kubernetes_state.deployment.count should contain only the DCA tags since it's executed on the cluster agent.
• check_run.kube_apiserver_controlplane.up.ok should contain both the DCA tags and Node Agent tags because the metric is produced from the kube_apiserver check that run on both the DCA and Node Agent.
• container.cpu.limit should contain just the Node Agent tags. This is a sanity check that extra tags aren't being applied in the wrong places.

DCA w/ CLC

1. Deploy the Agent. Uncomment out the KSM UseClusterCheckRunners and the clusterChecksRunners in the linux.yaml from above.
2. Deploy the nginx service from above.
3. Confirm the following metrics:

• network.http.cant_connect: This one will be different. The check is now being executed on the clusterChecksRunner which will not have the Node Agent tags. When the DCA creates and dispatches a check config it will attach its own global tags.
• kubernetes_state.deployment.count: DCA tags only
• check_run.kube_apiserver_controlplane.up.ok: DCA + Node tags
• container.cpu.limit: Node tags only

4. Confirm the following events tags:

Go to the Events Management Explorer and search for events created by your cluster. This will just be a lot of pod creations. Click on them and you should see all of the cluster-agent tags as well as the node-agent tags. We expect both because the orchestrator check runs on both the node-agent and the cluster-agent. They detected the same event and send data to the datadog backend with their own tags and they end up being merged together into one entity.

Possible Drawbacks / Trade-offs

Since these are low cardinality global tags, we don't expect to see a significant change in the number of unique metrics for customer billing.

This does in some ways limit the flexibility for some customers. For instance, they will not be able to use DD_ORCHESTRATOR_EXPLORER_EXTRA_TAGS to just add tags to the events emitted by the orchestrator check. Instead, the tag will be attached to all data from the cluster agent. This makes the tagging interface much simpler for customers.

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 create-vm --pipeline-id=49613788 --os-family=ubuntu

Note: This applies to commit 0e1801e

[cit-pr-commenter]

Regression Detector

Regression Detector Results

Metrics dashboard
Target profiles
Run ID: 1fe5c2f9-88e4-4785-a2b2-c3ff2107d6be

Baseline: a166214
Comparison: 0e1801e
Diff

Optimization Goals: ❌ Significant changes detected

perf	experiment	goal	Δ mean %	Δ mean % CI	trials	links
❌	pycheck_lots_of_tags	% cpu utilization	+8.86	[+5.23, +12.48]	1	Logs

Fine details of change detection per experiment

perf	experiment	goal	Δ mean %	Δ mean % CI	trials	links
❌	pycheck_lots_of_tags	% cpu utilization	+8.86	[+5.23, +12.48]	1	Logs
➖	basic_py_check	% cpu utilization	+4.66	[+0.80, +8.51]	1	Logs
➖	file_to_blackhole_500ms_latency	egress throughput	+0.26	[-0.51, +1.02]	1	Logs
➖	uds_dogstatsd_to_api_cpu	% cpu utilization	+0.11	[-0.61, +0.84]	1	Logs
➖	file_to_blackhole_0ms_latency	egress throughput	+0.11	[-0.78, +0.99]	1	Logs
➖	file_to_blackhole_1000ms_latency_linear_load	egress throughput	+0.10	[-0.36, +0.57]	1	Logs
➖	quality_gate_idle	memory utilization	+0.06	[+0.02, +0.11]	1	Logs bounds checks dashboard
➖	file_tree	memory utilization	+0.05	[-0.09, +0.20]	1	Logs
➖	file_to_blackhole_300ms_latency	egress throughput	+0.01	[-0.61, +0.64]	1	Logs
➖	tcp_dd_logs_filter_exclude	ingress throughput	-0.00	[-0.01, +0.01]	1	Logs
➖	uds_dogstatsd_to_api	ingress throughput	-0.00	[-0.11, +0.10]	1	Logs
➖	quality_gate_idle_all_features	memory utilization	-0.01	[-0.12, +0.10]	1	Logs bounds checks dashboard
➖	file_to_blackhole_100ms_latency	egress throughput	-0.04	[-0.76, +0.68]	1	Logs
➖	otel_to_otel_logs	ingress throughput	-0.24	[-0.94, +0.46]	1	Logs
➖	file_to_blackhole_1000ms_latency	egress throughput	-0.35	[-1.12, +0.42]	1	Logs
➖	tcp_syslog_to_blackhole	ingress throughput	-1.74	[-1.81, -1.66]	1	Logs

Bounds Checks: ❌ Failed

perf	experiment	bounds_check_name	replicates_passed	links
❌	file_to_blackhole_1000ms_latency	lost_bytes	0/10
✅	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_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	memory_usage	10/10	bounds checks dashboard
✅	quality_gate_idle_all_features	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_all_features, bounds check memory_usage: 10/10 replicas passed. Gate passed.
• quality_gate_idle, bounds check memory_usage: 10/10 replicas passed. Gate passed.

[gabedos]

Was testing the static/global tags by setting the same tag key to different values such as the following:

    DD_TAGS: "source:dca-tags"
    DD_EXTRA_TAGS: "source:dca-extra"
    DD_CLUSTER_CHECKS_EXTRA_TAGS: "source:dca-cluster_extra"
    DD_ORCHESTRATOR_EXPLORER_EXTRA_TAGS: "source:dca-orch_extra"

Realized only the last value was being recorded because this struct was not considering that the same tag key could have multiple values so change the type to map[string][]string

[gabedos]

Need to create a separate list of the global tags. This is because c.staticTags get applied directly to all of the containers and pods and we only want these global tags applied to the internal://global-entity-id. There will be some values duplicated in c.staticTags and globalEnvTags, however, the tagList removes duplicates and everything ends up being properly set on the GlobalEntity

[GustavoCaso]

@gabedos you are going to need to rebase|merge main on the PR. We recently changed the file structure for the tagger component and it will create merge conflicts #30584.

Let me know if you need any help 😄

[GustavoCaso]

LGTM!

Thanks for refactoring the global tags helper function and pass the config component as part of the function arguments

[gabedos]

/merge

[dd-devflow]

Devflow running: /merge

View all feedbacks in Devflow UI.

---

2024-11-20 18:25:18 UTC ℹ️ MergeQueue: waiting for PR to be ready

This merge request is not mergeable yet, because of pending checks/missing approvals. It will be added to the queue as soon as checks pass and/or get approvals.
Note: if you pushed new commits since the last approval, you may need additional approval.
You can remove it from the waiting list with /remove command.

---

2024-11-20 22:25:20 UTC ⚠️ MergeQueue: This merge request was unqueued

This merge request was unqueued

[gabedos]

/merge

[dd-devflow]

Devflow running: /merge

View all feedbacks in Devflow UI.

---

2024-11-21 18:36:19 UTC ℹ️ MergeQueue: pull request added to the queue

The median merge time in main is 25m.