[Backport 7.61.x] Performance improvements of External Metrics controller and allow multiple workers

[agent-platform-auto-pr]

Backport 5393674 from #31671.

---

What does this PR do?

Implement several improvements to allow much better scalability on External Metrics controller:

• Allow multiple workers through the external_metrics_provider.num_workers. With the new improvements, 2 workers can handle >3k DatadogMetric, although more workers could be necessary with a slow APIServer.
• Unlock store before doing APIServer Update calls
• Increase the limit of QPS/Burst on Kubernetes APIServer client (shared client) as we ultimately need to push 1 update for each DatadogMetric every 30s.
• Reduce the number of unnecessary

Motivation

Fix delayed/lagging Autoscaling on large deployments.

Describe how you validated your changes

The performance improvement can only be seen on large clusters with a lot of DatadogMetric objects (starting ~600).
Without this PR, a degradation in the frequency and freshness of data visible in the DatadogMetric objects can be seen.
With this PR, the updates should be available on time every 30s.

Outside of that, only non-regression

Possible Drawbacks / Trade-offs

Additional Notes

If the controller is already in a lagging state, the amount of requests to APIServer is going to increase a lot.

[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=50316177 --os-family=ubuntu

Note: This applies to commit 8570bf7

[cit-pr-commenter]

Regression Detector

Regression Detector Results

Metrics dashboard
Target profiles
Run ID: ee9580bd-2a24-4e0f-8028-fc009a762bd8

Baseline: 273b005
Comparison: 8570bf7
Diff

Optimization Goals: ❌ Regression(s) detected

perf	experiment	goal	Δ mean %	Δ mean % CI	trials	links
❌	basic_py_check	% cpu utilization	+5.73	[+1.64, +9.81]	1	Logs

Fine details of change detection per experiment

perf	experiment	goal	Δ mean %	Δ mean % CI	trials	links
❌	basic_py_check	% cpu utilization	+5.73	[+1.64, +9.81]	1	Logs
➖	otel_to_otel_logs	ingress throughput	+1.60	[+0.85, +2.35]	1	Logs
➖	pycheck_lots_of_tags	% cpu utilization	+1.03	[-2.35, +4.40]	1	Logs
➖	file_to_blackhole_1000ms_latency_linear_load	egress throughput	+0.14	[-0.33, +0.61]	1	Logs
➖	file_to_blackhole_100ms_latency	egress throughput	+0.12	[-0.60, +0.85]	1	Logs
➖	file_to_blackhole_300ms_latency	egress throughput	+0.12	[-0.51, +0.75]	1	Logs
➖	file_to_blackhole_500ms_latency	egress throughput	+0.09	[-0.70, +0.87]	1	Logs
➖	file_to_blackhole_0ms_latency	egress throughput	+0.08	[-0.77, +0.93]	1	Logs
➖	tcp_dd_logs_filter_exclude	ingress throughput	+0.00	[-0.01, +0.01]	1	Logs
➖	uds_dogstatsd_to_api	ingress throughput	-0.01	[-0.11, +0.09]	1	Logs
➖	quality_gate_idle_all_features	memory utilization	-0.07	[-0.18, +0.04]	1	Logs bounds checks dashboard
➖	tcp_syslog_to_blackhole	ingress throughput	-0.17	[-0.23, -0.11]	1	Logs
➖	quality_gate_idle	memory utilization	-0.21	[-0.25, -0.17]	1	Logs bounds checks dashboard
➖	file_to_blackhole_1000ms_latency	egress throughput	-0.26	[-1.02, +0.50]	1	Logs
➖	file_tree	memory utilization	-0.32	[-0.47, -0.17]	1	Logs
➖	uds_dogstatsd_to_api_cpu	% cpu utilization	-0.74	[-1.46, -0.02]	1	Logs
➖	quality_gate_logs	% cpu utilization	-1.35	[-4.26, +1.55]	1	Logs

Bounds Checks: ❌ Failed

perf	experiment	bounds_check_name	replicates_passed	links
❌	file_to_blackhole_100ms_latency	lost_bytes	9/10
✅	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	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
✅	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_idle_all_features, bounds check memory_usage: 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.

[vboulineau]

/merge

[dd-devflow]

Devflow running: /merge

View all feedbacks in Devflow UI.

---

2024-12-04 10:33:45 UTC ℹ️ MergeQueue: pull request added to the queue

The median merge time in 7.61.x is 24m.