Add k6 activity ramp benchmark

[rgarcia]

Summary

• add a k6 TypeScript activity-ramp benchmark for Hypeman instance lifecycle load
• add noob-friendly comments explaining the k6 execution model, VUs, scenarios, metrics, setup, teardown, and activity loop
• move the benchmark make target into benchmarks/Makefile and document running it with make -C benchmarks
• add hypervisor selection and tagging for cloud-hypervisor, firecracker, and qemu matrix runs
• make benchmark setup compatible with image names containing slashes and preserve unique instance suffixes under long run IDs
• record create admission rejections separately with a short backoff so capacity limits do not look like script failures

Validation

• make -C benchmarks -n bench-activity-ramp HYPEMAN_API_KEY=dummy
• docker run --rm -v "$PWD":/src -w /src/benchmarks grafana/k6:latest inspect k6/activity-ramp.ts
• git diff --check origin/main...HEAD
• previously ran 24 matrix benchmark runs across m8i host sizes, gp3 settings, and cloud-hypervisor/firecracker/qemu
• previously verified no active benchmark stacks remained after the matrix

CloudFormation quickstart options for ingress and gp3 tuning live separately in PR #237.

---

Note

Low Risk
Low risk: changes are additive benchmark tooling/docs plus ignoring local .bench/ artifacts, with no runtime or production code modifications.

Overview
Adds a new k6 TypeScript benchmark (benchmarks/k6/activity-ramp.ts) that ramps VUs running an instance lifecycle loop (create → wait for Running → HTTP probe via shared pattern ingress → delete), including setup/teardown cleanup and metrics for latency, success rates, and capacity rejections (409s) with backoff.

Introduces benchmarks/Makefile to run the benchmark via make -C benchmarks bench-activity-ramp, exporting HTML/JSON reports under .bench/k6, and documents usage/tuning in benchmarks/k6/README.md. Also updates .gitignore to exclude .bench/ output.

^{Reviewed by Cursor Bugbot for commit 8336845. Bugbot is set up for automated code reviews on this repo. Configure here.}

[firetiger-agent]

Firetiger deploy monitoring skipped

This PR didn't match the auto-monitor filter configured on your GitHub connection:

Any PR that changes the kernel API. Monitor changes to API endpoints (packages/api/cmd/api/) and Temporal workflows (packages/api/lib/temporal) in the kernel repo

Reason: PR adds k6 benchmarking infrastructure in the benchmarks directory, not changes to API endpoints (packages/api/cmd/api/) or Temporal workflows (packages/api/lib/temporal).

To monitor this PR anyway, reply with @firetiger monitor this.

[cursor]

Cursor Bugbot has reviewed your changes and found 2 potential issues.

Autofix Details

Bugbot Autofix prepared fixes for both issues found in the latest run.

• ✅ Fixed: Ingress host port default mismatch between Makefile and TypeScript
  • Updated the loadConfig() fallback for HYPEMAN_INGRESS_HOST_PORT from 8081 to 80 so direct k6 runs match Makefile and README defaults.
• ✅ Fixed: Instance body tags use per-VU config.runId instead of shared data.runId
  • Changed instance creation tags to use tags.run_id so created instances share the setup-provided run ID that teardown queries for cleanup.

Or push these changes by commenting:

@cursor push a5823bc801

Preview (a5823bc801)

diff --git a/benchmarks/k6/activity-ramp.ts b/benchmarks/k6/activity-ramp.ts
--- a/benchmarks/k6/activity-ramp.ts
+++ b/benchmarks/k6/activity-ramp.ts
@@ -139,7 +139,7 @@
 
 function loadConfig(): Config {
   const baseUrl = trimRight(requiredEnv('HYPEMAN_BASE_URL', 'http://127.0.0.1:8080'), '/');
-  const ingressHostPort = intEnv('HYPEMAN_INGRESS_HOST_PORT', 8081);
+  const ingressHostPort = intEnv('HYPEMAN_INGRESS_HOST_PORT', 80);
 
   return {
     baseUrl,
@@ -297,7 +297,7 @@
     tags: {
       benchmark: 'activity-ramp',
       hypervisor: config.hypervisor || 'server-default',
-      run_id: config.runId,
+      run_id: tags.run_id,
     },
     skip_kernel_headers: true,
   };

_{You can send follow-ups to the cloud agent here.}

[cursor]

Cursor Bugbot has reviewed your changes and found 1 potential issue.

Bugbot Autofix prepared a fix for the issue found in the latest run.

• ✅ Fixed: requiredEnv is identical to envString, adds no validation
  • requiredEnv now fails when the resolved value is empty, so missing required variables with empty fallback are enforced instead of silently defaulting.

Or push these changes by commenting:

@cursor push cb21feedc7

Preview (cb21feedc7)

diff --git a/benchmarks/k6/activity-ramp.ts b/benchmarks/k6/activity-ramp.ts
--- a/benchmarks/k6/activity-ramp.ts
+++ b/benchmarks/k6/activity-ramp.ts
@@ -492,7 +492,11 @@
 }
 
 function requiredEnv(name: string, fallback: string): string {
-  return envString(name, fallback);
+  const value = envString(name, fallback);
+  if (value === '') {
+    fail(`${name} is required`);
+  }
+  return value;
 }
 
 function intEnv(name: string, fallback: number): number {

_{You can send follow-ups to the cloud agent here.}

^{Reviewed by Cursor Bugbot for commit 8336845. Configure here.}

[cursor]

requiredEnv is identical to envString, adds no validation

Low Severity

requiredEnv is a direct pass-through to envString with no additional validation — the function body is just return envString(name, fallback). The "required" semantics are misleading: two of its three call sites (HYPEMAN_BASE_URL, HYPEMAN_IMAGE) supply non-empty fallbacks, so the variables are never truly required. Actual validation of HYPEMAN_API_KEY happens separately in checkRequiredConfig. A future contributor who calls requiredEnv('NEW_VAR', '') expecting it to fail on a missing value would get a silent empty-string default instead.

Additional Locations (1)

• benchmarks/k6/activity-ramp.ts#L140-L147

 

^{Reviewed by Cursor Bugbot for commit 8336845. Configure here.}