[CSM] Fix syscall based drift events

[Gui774ume]

What does this PR do?

This PR improves how we generate drift events for syscalls past the initial activity dump learning period. Leveraging the same aggregation / conditional send logique, we're now leveraging the same event type for both the syscalls collection events during the activity dump learning period, and the syscall based drift events.

Motivation

This change is necessary to make syscall drift events work while ensuring we won't spam events because of poorly profiled workloads.

Additional Notes

Although the list of syscalls is collected and recorded in dumps per process, their evaluation is based on a flattened list that contains all the syscalls made across all processes.

Describe how to test/QA your changes

• Load a minimalistic profile, turn on drift events for syscalls and make sure a syscall anomaly detection event is sent:
  • after a new process exits
  • after a new process execve into something else
  • after 1 minute of making new (at least one new) or existing syscalls in the profile

[pr-commenter]

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=34718488 --os-family=ubuntu

[pr-commenter]

Regression Detector

Regression Detector Results

Run ID: 7e7c1346-a045-42fd-8fe2-10dc0eb209c4
Baseline: bbfeda2
Comparison: 9a96bf5

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

No significant changes in experiment optimization goals

Confidence level: 90.00%
Effect size tolerance: |Δ mean %| ≥ 5.00%

There were no significant changes in experiment optimization goals at this confidence level and effect size tolerance.

Fine details of change detection per experiment

perf	experiment	goal	Δ mean %	Δ mean % CI
➖	tcp_syslog_to_blackhole	ingress throughput	+2.47	[-18.59, +23.53]
➖	pycheck_1000_100byte_tags	% cpu utilization	+0.59	[-4.12, +5.29]
➖	trace_agent_msgpack	ingress throughput	+0.01	[+0.00, +0.02]
➖	uds_dogstatsd_to_api	ingress throughput	+0.00	[-0.20, +0.20]
➖	trace_agent_json	ingress throughput	-0.00	[-0.03, +0.03]
➖	tcp_dd_logs_filter_exclude	ingress throughput	-0.00	[-0.05, +0.04]
➖	otel_to_otel_logs	ingress throughput	-0.08	[-0.43, +0.28]
➖	basic_py_check	% cpu utilization	-0.24	[-2.73, +2.26]
➖	idle	memory utilization	-0.25	[-0.29, -0.20]
➖	file_tree	memory utilization	-0.67	[-0.80, -0.54]
➖	uds_dogstatsd_to_api_cpu	% cpu utilization	-2.04	[-4.84, +0.76]

Explanation

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".

[cswatt]

approving for documentation

[Gui774ume]

/merge

[dd-devflow]

🚂 MergeQueue

Pull request added to the queue.

There are 2 builds ahead! (estimated merge in less than 1h)

Use /merge -c to cancel this operation!