enhancement(sample transform): add stratified sampling capability

[hillmandj]

Resolves issue. The goal of this PR is to leverage the use of a HashMap to attempt to enable stratified sampling. The idea is that a user can specify a field in a sampler, such as: service, and each event with a unique service will then be sampled together. The motivation behind this is that currently we need to create individual samplers with basically the same configuration over and over in order to achieve the same rate for a particular log stream.

Should the option not be passed, the log event should fallback to the None key in the hashmap, and act as a regular counter just as we had before.

A note about this PR/me: I am not a rust developer. This is my first attempt at doing anything with respect to the rust programming language. I am likely doing something wrong (not typing properly, missing edge cases, not testing properly, etc.) and am looking for feedback as to how to make this better / more idiomatic. Hopefully maintainers / actual rust experts will see the motivation and give me some direction in order to make this mergeable. Thanks! 🎉

[bits-bot]

All committers have signed the CLA.

[hillmandj]

@jszwedko I have just accepted the CLA. Please let me know if there's anything else I need to do to get a review on the PR. Thanks!

[jszwedko]

@jszwedko I have just accepted the CLA. Please let me know if there's anything else I need to do to get a review on the PR. Thanks!

Thank you! We'll take a look at this shortly.

[hillmandj]

Tagging @pront as well, hoping to get feedback from ya'll soon!

[jszwedko]

Apologies for the delay in review of this @hillmandj ! This is a valuable enhancement.

I left a few inline comments, but I think this is mostly there. A couple of other notes:

• Could you add a changelog entry? See https://github.com/vectordotdev/vector/blob/master/changelog.d/README.md for a description of the format.
• Could you run make generate-component-docs? That should regenerate the documentation with the new field you added.

[jszwedko]

Awesome, thanks for this contribution @hillmandj !

[hillmandj]

@jszwedko thank you so much for your feedback! I hope that I addressed all of your concerns. Please let me know if you think any more changes are necessary, or if there is some manual testing I should do to ensure this works out before merging. Appreciate your help and happy to contribute! 🎉

[maycmlee]

Just a small edit!

[maycmlee]

Suggested change

	If left unspecified, or if the event doesn't have `group_by`, then the event is not sampled
	If left unspecified, or if the event doesn't have `group_by`, then the event is not sampled.

[hillmandj]

@jszwedko pushed up some small formatting changes after seeing the output of this failed check, and also attempted to address @maycmlee's comment.

cargo vdev check fmt succeeds locally 👍

[github-actions]

Regression Detector Results

Run ID: a73d394e-907e-41fc-98e1-74e265e19d6c Metrics dashboard

Baseline: 28060db
Comparison: 8fbb355

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

Significant changes in experiment optimization goals

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

perf	experiment	goal	Δ mean %	Δ mean % CI	links
❌	otlp_http_to_blackhole	ingress throughput	-5.64	[-5.78, -5.51]

Experiments ignored for regressions

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

perf	experiment	goal	Δ mean %	Δ mean % CI	links
➖	file_to_blackhole	egress throughput	-5.78	[-12.42, +0.86]

Fine details of change detection per experiment

perf	experiment	goal	Δ mean %	Δ mean % CI	links
➖	syslog_log2metric_humio_metrics	ingress throughput	+2.38	[+2.23, +2.54]
➖	datadog_agent_remap_datadog_logs	ingress throughput	+2.18	[+1.97, +2.40]
➖	syslog_log2metric_tag_cardinality_limit_blackhole	ingress throughput	+1.78	[+1.68, +1.88]
➖	http_to_http_acks	ingress throughput	+1.44	[+0.22, +2.67]
➖	http_elasticsearch	ingress throughput	+1.11	[+0.89, +1.32]
➖	syslog_regex_logs2metric_ddmetrics	ingress throughput	+1.09	[+0.93, +1.26]
➖	syslog_log2metric_splunk_hec_metrics	ingress throughput	+1.03	[+0.92, +1.13]
➖	syslog_splunk_hec_logs	ingress throughput	+0.60	[+0.51, +0.69]
➖	datadog_agent_remap_blackhole	ingress throughput	+0.22	[+0.10, +0.34]
➖	http_to_http_noack	ingress throughput	+0.19	[+0.10, +0.29]
➖	http_to_http_json	ingress throughput	+0.03	[-0.01, +0.07]
➖	splunk_hec_to_splunk_hec_logs_noack	ingress throughput	+0.02	[-0.08, +0.12]
➖	splunk_hec_indexer_ack_blackhole	ingress throughput	+0.02	[-0.07, +0.11]
➖	http_to_s3	ingress throughput	+0.02	[-0.25, +0.28]
➖	splunk_hec_to_splunk_hec_logs_acks	ingress throughput	+0.00	[-0.10, +0.11]
➖	otlp_grpc_to_blackhole	ingress throughput	-0.81	[-0.94, -0.68]
➖	http_text_to_http_json	ingress throughput	-1.30	[-1.43, -1.17]
➖	fluent_elasticsearch	ingress throughput	-1.35	[-1.84, -0.86]
➖	syslog_loki	ingress throughput	-1.50	[-1.57, -1.42]
➖	datadog_agent_remap_blackhole_acks	ingress throughput	-1.51	[-1.62, -1.40]
➖	splunk_hec_route_s3	ingress throughput	-1.52	[-1.84, -1.20]
➖	datadog_agent_remap_datadog_logs_acks	ingress throughput	-1.63	[-1.82, -1.43]
➖	syslog_humio_logs	ingress throughput	-2.34	[-2.45, -2.23]
➖	socket_to_socket_blackhole	ingress throughput	-4.34	[-4.41, -4.28]
❌	otlp_http_to_blackhole	ingress throughput	-5.64	[-5.78, -5.51]
➖	file_to_blackhole	egress throughput	-5.78	[-12.42, +0.86]

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".

[github-actions]

Regression Detector Results

Run ID: c112d4a6-038e-4b42-a7da-1794d3da2b93 Metrics dashboard

Baseline: 7feb20b
Comparison: 1bdb614

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

Significant changes in experiment optimization goals

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

perf	experiment	goal	Δ mean %	Δ mean % CI	links
✅	socket_to_socket_blackhole	ingress throughput	+6.14	[+6.05, +6.23]

Experiments ignored for regressions

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

perf	experiment	goal	Δ mean %	Δ mean % CI	links
❌	file_to_blackhole	egress throughput	-11.43	[-17.85, -5.00]

Fine details of change detection per experiment

perf	experiment	goal	Δ mean %	Δ mean % CI	links
✅	socket_to_socket_blackhole	ingress throughput	+6.14	[+6.05, +6.23]
➖	otlp_http_to_blackhole	ingress throughput	+3.86	[+3.74, +3.97]
➖	syslog_loki	ingress throughput	+1.86	[+1.77, +1.95]
➖	datadog_agent_remap_datadog_logs_acks	ingress throughput	+1.55	[+1.37, +1.73]
➖	syslog_log2metric_splunk_hec_metrics	ingress throughput	+1.51	[+1.40, +1.62]
➖	otlp_grpc_to_blackhole	ingress throughput	+0.98	[+0.87, +1.09]
➖	datadog_agent_remap_blackhole	ingress throughput	+0.65	[+0.52, +0.78]
➖	http_to_http_acks	ingress throughput	+0.53	[-0.68, +1.75]
➖	http_to_http_json	ingress throughput	+0.04	[+0.00, +0.08]
➖	http_to_http_noack	ingress throughput	+0.02	[-0.02, +0.05]
➖	splunk_hec_to_splunk_hec_logs_noack	ingress throughput	+0.01	[-0.08, +0.11]
➖	splunk_hec_to_splunk_hec_logs_acks	ingress throughput	+0.00	[-0.10, +0.11]
➖	splunk_hec_indexer_ack_blackhole	ingress throughput	-0.00	[-0.08, +0.08]
➖	http_to_s3	ingress throughput	-0.21	[-0.48, +0.05]
➖	http_elasticsearch	ingress throughput	-0.65	[-0.81, -0.48]
➖	syslog_humio_logs	ingress throughput	-0.66	[-0.77, -0.55]
➖	splunk_hec_route_s3	ingress throughput	-0.75	[-1.06, -0.43]
➖	datadog_agent_remap_datadog_logs	ingress throughput	-0.86	[-1.07, -0.65]
➖	syslog_splunk_hec_logs	ingress throughput	-1.12	[-1.24, -0.99]
➖	datadog_agent_remap_blackhole_acks	ingress throughput	-1.16	[-1.27, -1.05]
➖	syslog_log2metric_humio_metrics	ingress throughput	-1.37	[-1.53, -1.20]
➖	fluent_elasticsearch	ingress throughput	-1.45	[-1.93, -0.96]
➖	http_text_to_http_json	ingress throughput	-2.28	[-2.42, -2.13]
➖	syslog_log2metric_tag_cardinality_limit_blackhole	ingress throughput	-2.53	[-2.73, -2.33]
➖	syslog_regex_logs2metric_ddmetrics	ingress throughput	-3.84	[-3.96, -3.71]
❌	file_to_blackhole	egress throughput	-11.43	[-17.85, -5.00]

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".

[hillmandj]

@jszwedko @pront looks like the PR got removed from the merge queue due to this failing check, which seems unrelated to my change. It appears the job is failing to install the awscli ubuntu package. Hoping this is just a random hiccup, but please let me know if there is anything I can do to help! 🙏

[github-actions]

Regression Detector Results

Run ID: 62869838-49ba-4e91-a137-032d5688f9e1 Metrics dashboard

Baseline: 8385db5
Comparison: f741170

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

Significant changes in experiment optimization goals

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

perf	experiment	goal	Δ mean %	Δ mean % CI	links
✅	socket_to_socket_blackhole	ingress throughput	+5.91	[+5.84, +5.98]
✅	otlp_http_to_blackhole	ingress throughput	+5.10	[+5.01, +5.18]

Experiments ignored for regressions

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

perf	experiment	goal	Δ mean %	Δ mean % CI	links
➖	file_to_blackhole	egress throughput	-3.58	[-7.55, +0.39]

Fine details of change detection per experiment

perf	experiment	goal	Δ mean %	Δ mean % CI	links
✅	socket_to_socket_blackhole	ingress throughput	+5.91	[+5.84, +5.98]
✅	otlp_http_to_blackhole	ingress throughput	+5.10	[+5.01, +5.18]
➖	syslog_log2metric_humio_metrics	ingress throughput	+2.40	[+2.23, +2.56]
➖	syslog_loki	ingress throughput	+2.40	[+2.33, +2.46]
➖	datadog_agent_remap_datadog_logs_acks	ingress throughput	+2.29	[+2.16, +2.42]
➖	splunk_hec_route_s3	ingress throughput	+0.90	[+0.59, +1.21]
➖	datadog_agent_remap_blackhole	ingress throughput	+0.61	[+0.53, +0.68]
➖	http_to_s3	ingress throughput	+0.10	[-0.06, +0.25]
➖	syslog_log2metric_splunk_hec_metrics	ingress throughput	+0.09	[-0.02, +0.20]
➖	syslog_humio_logs	ingress throughput	+0.08	[-0.01, +0.17]
➖	syslog_splunk_hec_logs	ingress throughput	+0.04	[-0.04, +0.12]
➖	fluent_elasticsearch	ingress throughput	+0.04	[-0.49, +0.56]
➖	http_to_http_acks	ingress throughput	+0.00	[-0.48, +0.49]
➖	splunk_hec_indexer_ack_blackhole	ingress throughput	+0.00	[-0.02, +0.02]
➖	http_to_http_noack	ingress throughput	+0.00	[-0.05, +0.06]
➖	splunk_hec_to_splunk_hec_logs_noack	ingress throughput	-0.00	[-0.02, +0.02]
➖	splunk_hec_to_splunk_hec_logs_acks	ingress throughput	-0.00	[-0.02, +0.02]
➖	http_to_http_json	ingress throughput	-0.01	[-0.05, +0.04]
➖	http_text_to_http_json	ingress throughput	-0.23	[-0.38, -0.08]
➖	datadog_agent_remap_blackhole_acks	ingress throughput	-0.72	[-0.81, -0.62]
➖	http_elasticsearch	ingress throughput	-0.85	[-0.97, -0.73]
➖	datadog_agent_remap_datadog_logs	ingress throughput	-0.86	[-1.01, -0.70]
➖	otlp_grpc_to_blackhole	ingress throughput	-2.84	[-3.02, -2.66]
➖	syslog_regex_logs2metric_ddmetrics	ingress throughput	-3.07	[-3.29, -2.85]
➖	syslog_log2metric_tag_cardinality_limit_blackhole	ingress throughput	-3.22	[-3.36, -3.09]
➖	file_to_blackhole	egress throughput	-3.58	[-7.55, +0.39]

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".

[github-actions]

Regression Detector Results

Run ID: fc44e88a-a59e-447e-8a5d-6dd6b3c77b1d Metrics dashboard

Baseline: d2f855c
Comparison: 7f379b9

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

Significant changes in experiment optimization goals

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

perf	experiment	goal	Δ mean %	Δ mean % CI	links
❌	otlp_http_to_blackhole	ingress throughput	-5.63	[-5.71, -5.55]

Experiments ignored for regressions

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

perf	experiment	goal	Δ mean %	Δ mean % CI	links
➖	file_to_blackhole	egress throughput	-0.10	[-4.28, +4.08]

Fine details of change detection per experiment

perf	experiment	goal	Δ mean %	Δ mean % CI	links
➖	datadog_agent_remap_datadog_logs_acks	ingress throughput	+3.13	[+2.95, +3.31]
➖	datadog_agent_remap_blackhole_acks	ingress throughput	+2.33	[+2.22, +2.43]
➖	datadog_agent_remap_datadog_logs	ingress throughput	+2.29	[+2.12, +2.46]
➖	http_elasticsearch	ingress throughput	+1.52	[+1.34, +1.69]
➖	syslog_log2metric_tag_cardinality_limit_blackhole	ingress throughput	+0.78	[+0.69, +0.88]
➖	datadog_agent_remap_blackhole	ingress throughput	+0.70	[+0.60, +0.80]
➖	splunk_hec_route_s3	ingress throughput	+0.65	[+0.34, +0.97]
➖	otlp_grpc_to_blackhole	ingress throughput	+0.53	[+0.33, +0.72]
➖	http_to_s3	ingress throughput	+0.10	[-0.06, +0.27]
➖	splunk_hec_to_splunk_hec_logs_noack	ingress throughput	-0.00	[-0.02, +0.02]
➖	splunk_hec_to_splunk_hec_logs_acks	ingress throughput	-0.00	[-0.03, +0.03]
➖	splunk_hec_indexer_ack_blackhole	ingress throughput	-0.00	[-0.02, +0.02]
➖	http_to_http_noack	ingress throughput	-0.01	[-0.07, +0.05]
➖	http_to_http_json	ingress throughput	-0.01	[-0.06, +0.04]
➖	http_to_http_acks	ingress throughput	-0.03	[-0.52, +0.46]
➖	file_to_blackhole	egress throughput	-0.10	[-4.28, +4.08]
➖	syslog_loki	ingress throughput	-0.59	[-0.68, -0.51]
➖	syslog_log2metric_humio_metrics	ingress throughput	-0.64	[-0.80, -0.49]
➖	http_text_to_http_json	ingress throughput	-0.72	[-0.86, -0.58]
➖	fluent_elasticsearch	ingress throughput	-0.89	[-1.42, -0.37]
➖	syslog_humio_logs	ingress throughput	-2.54	[-2.67, -2.41]
➖	syslog_regex_logs2metric_ddmetrics	ingress throughput	-2.59	[-2.77, -2.41]
➖	syslog_log2metric_splunk_hec_metrics	ingress throughput	-2.70	[-2.77, -2.62]
➖	syslog_splunk_hec_logs	ingress throughput	-3.79	[-3.88, -3.69]
➖	socket_to_socket_blackhole	ingress throughput	-4.68	[-4.76, -4.61]
❌	otlp_http_to_blackhole	ingress throughput	-5.63	[-5.71, -5.55]

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".

[github-actions]

Regression Detector Results

Run ID: 8e1729c0-1696-420a-ad72-5972e88d3ea5 Metrics dashboard

Baseline: 7054ae0
Comparison: 3baebb4

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

Significant changes in experiment optimization goals

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

perf	experiment	goal	Δ mean %	Δ mean % CI	links
✅	socket_to_socket_blackhole	ingress throughput	+6.03	[+5.98, +6.09]

Experiments ignored for regressions

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

perf	experiment	goal	Δ mean %	Δ mean % CI	links
➖	file_to_blackhole	egress throughput	-0.42	[-4.55, +3.71]

Fine details of change detection per experiment

perf	experiment	goal	Δ mean %	Δ mean % CI	links
✅	socket_to_socket_blackhole	ingress throughput	+6.03	[+5.98, +6.09]
➖	syslog_humio_logs	ingress throughput	+4.06	[+3.93, +4.20]
➖	http_text_to_http_json	ingress throughput	+3.90	[+3.77, +4.03]
➖	http_elasticsearch	ingress throughput	+2.93	[+2.76, +3.10]
➖	otlp_http_to_blackhole	ingress throughput	+2.79	[+2.70, +2.89]
➖	datadog_agent_remap_blackhole_acks	ingress throughput	+2.51	[+2.41, +2.61]
➖	syslog_log2metric_humio_metrics	ingress throughput	+1.76	[+1.60, +1.93]
➖	syslog_regex_logs2metric_ddmetrics	ingress throughput	+1.67	[+1.53, +1.82]
➖	datadog_agent_remap_datadog_logs	ingress throughput	+0.76	[+0.57, +0.95]
➖	fluent_elasticsearch	ingress throughput	+0.63	[+0.08, +1.17]
➖	syslog_log2metric_tag_cardinality_limit_blackhole	ingress throughput	+0.42	[+0.35, +0.50]
➖	syslog_splunk_hec_logs	ingress throughput	+0.37	[+0.27, +0.47]
➖	http_to_http_json	ingress throughput	+0.00	[-0.04, +0.05]
➖	splunk_hec_indexer_ack_blackhole	ingress throughput	+0.00	[-0.02, +0.02]
➖	splunk_hec_to_splunk_hec_logs_noack	ingress throughput	+0.00	[-0.02, +0.02]
➖	splunk_hec_to_splunk_hec_logs_acks	ingress throughput	-0.01	[-0.04, +0.02]
➖	http_to_http_noack	ingress throughput	-0.01	[-0.07, +0.05]
➖	http_to_http_acks	ingress throughput	-0.07	[-0.56, +0.42]
➖	splunk_hec_route_s3	ingress throughput	-0.42	[-0.71, -0.13]
➖	file_to_blackhole	egress throughput	-0.42	[-4.55, +3.71]
➖	http_to_s3	ingress throughput	-0.61	[-0.78, -0.44]
➖	syslog_log2metric_splunk_hec_metrics	ingress throughput	-0.66	[-0.79, -0.54]
➖	datadog_agent_remap_datadog_logs_acks	ingress throughput	-0.98	[-1.17, -0.79]
➖	datadog_agent_remap_blackhole	ingress throughput	-1.02	[-1.11, -0.93]
➖	syslog_loki	ingress throughput	-1.17	[-1.21, -1.12]
➖	otlp_grpc_to_blackhole	ingress throughput	-3.61	[-3.79, -3.42]

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".

[github-actions]

Regression Detector Results

Run ID: 26932709-634d-45a6-85a5-f0b7ca6f7a48 Metrics dashboard

Baseline: 650e6b2
Comparison: 11af229

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

Significant changes in experiment optimization goals

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

perf	experiment	goal	Δ mean %	Δ mean % CI	links
❌	otlp_http_to_blackhole	ingress throughput	-6.33	[-6.42, -6.24]

Experiments ignored for regressions

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

perf	experiment	goal	Δ mean %	Δ mean % CI	links
➖	file_to_blackhole	egress throughput	-2.34	[-6.40, +1.72]

Fine details of change detection per experiment

perf	experiment	goal	Δ mean %	Δ mean % CI	links
➖	otlp_grpc_to_blackhole	ingress throughput	+4.83	[+4.63, +5.02]
➖	syslog_log2metric_humio_metrics	ingress throughput	+2.53	[+2.41, +2.65]
➖	datadog_agent_remap_datadog_logs_acks	ingress throughput	+1.81	[+1.69, +1.94]
➖	syslog_splunk_hec_logs	ingress throughput	+1.53	[+1.45, +1.62]
➖	syslog_log2metric_splunk_hec_metrics	ingress throughput	+0.69	[+0.60, +0.79]
➖	http_to_s3	ingress throughput	+0.66	[+0.50, +0.82]
➖	datadog_agent_remap_blackhole	ingress throughput	+0.45	[+0.38, +0.53]
➖	fluent_elasticsearch	ingress throughput	+0.18	[-0.35, +0.70]
➖	datadog_agent_remap_datadog_logs	ingress throughput	+0.15	[-0.06, +0.36]
➖	syslog_loki	ingress throughput	+0.02	[-0.05, +0.09]
➖	http_to_http_acks	ingress throughput	+0.02	[-0.47, +0.50]
➖	http_to_http_json	ingress throughput	+0.01	[-0.04, +0.06]
➖	splunk_hec_indexer_ack_blackhole	ingress throughput	+0.00	[-0.02, +0.02]
➖	splunk_hec_to_splunk_hec_logs_noack	ingress throughput	-0.00	[-0.02, +0.02]
➖	splunk_hec_to_splunk_hec_logs_acks	ingress throughput	-0.00	[-0.02, +0.02]
➖	http_to_http_noack	ingress throughput	-0.00	[-0.06, +0.05]
➖	syslog_humio_logs	ingress throughput	-0.00	[-0.11, +0.11]
➖	datadog_agent_remap_blackhole_acks	ingress throughput	-0.52	[-0.61, -0.43]
➖	syslog_regex_logs2metric_ddmetrics	ingress throughput	-0.62	[-0.76, -0.48]
➖	syslog_log2metric_tag_cardinality_limit_blackhole	ingress throughput	-0.97	[-1.06, -0.88]
➖	http_elasticsearch	ingress throughput	-1.05	[-1.22, -0.88]
➖	splunk_hec_route_s3	ingress throughput	-1.43	[-1.75, -1.10]
➖	http_text_to_http_json	ingress throughput	-1.78	[-1.90, -1.65]
➖	file_to_blackhole	egress throughput	-2.34	[-6.40, +1.72]
➖	socket_to_socket_blackhole	ingress throughput	-4.48	[-4.54, -4.41]
❌	otlp_http_to_blackhole	ingress throughput	-6.33	[-6.42, -6.24]

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".

[github-actions]

Regression Detector Results

Run ID: b1feac77-098d-4ec3-a3e1-cca691f687fd Metrics dashboard

Baseline: 624fa1a
Comparison: 448ce64

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.

Experiments ignored for regressions

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

perf	experiment	goal	Δ mean %	Δ mean % CI	links
➖	file_to_blackhole	egress throughput	-1.11	[-5.23, +3.00]

Fine details of change detection per experiment

perf	experiment	goal	Δ mean %	Δ mean % CI	links
➖	syslog_log2metric_splunk_hec_metrics	ingress throughput	+3.52	[+3.37, +3.67]
➖	syslog_log2metric_humio_metrics	ingress throughput	+1.76	[+1.61, +1.91]
➖	syslog_log2metric_tag_cardinality_limit_blackhole	ingress throughput	+1.37	[+1.26, +1.48]
➖	syslog_humio_logs	ingress throughput	+1.03	[+0.94, +1.13]
➖	http_elasticsearch	ingress throughput	+0.94	[+0.76, +1.12]
➖	syslog_loki	ingress throughput	+0.91	[+0.82, +1.00]
➖	http_to_s3	ingress throughput	+0.68	[+0.50, +0.86]
➖	datadog_agent_remap_blackhole	ingress throughput	+0.47	[+0.38, +0.55]
➖	datadog_agent_remap_datadog_logs_acks	ingress throughput	+0.43	[+0.30, +0.56]
➖	syslog_splunk_hec_logs	ingress throughput	+0.22	[+0.12, +0.31]
➖	socket_to_socket_blackhole	ingress throughput	+0.05	[-0.02, +0.12]
➖	http_to_http_acks	ingress throughput	+0.04	[-0.46, +0.53]
➖	http_to_http_noack	ingress throughput	+0.01	[-0.04, +0.07]
➖	http_text_to_http_json	ingress throughput	-0.00	[-0.14, +0.14]
➖	splunk_hec_to_splunk_hec_logs_noack	ingress throughput	-0.00	[-0.02, +0.02]
➖	splunk_hec_indexer_ack_blackhole	ingress throughput	-0.00	[-0.02, +0.02]
➖	http_to_http_json	ingress throughput	-0.00	[-0.04, +0.04]
➖	splunk_hec_to_splunk_hec_logs_acks	ingress throughput	-0.00	[-0.03, +0.03]
➖	splunk_hec_route_s3	ingress throughput	-0.05	[-0.35, +0.24]
➖	datadog_agent_remap_datadog_logs	ingress throughput	-0.13	[-0.34, +0.07]
➖	otlp_grpc_to_blackhole	ingress throughput	-0.21	[-0.43, +0.00]
➖	syslog_regex_logs2metric_ddmetrics	ingress throughput	-0.46	[-0.61, -0.31]
➖	file_to_blackhole	egress throughput	-1.11	[-5.23, +3.00]
➖	datadog_agent_remap_blackhole_acks	ingress throughput	-1.23	[-1.34, -1.12]
➖	otlp_http_to_blackhole	ingress throughput	-1.25	[-1.33, -1.16]
➖	fluent_elasticsearch	ingress throughput	-2.53	[-3.06, -2.00]

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".