Lightweight performance tests that measure baseline latencies for core HyperFleet operations. Tests run inside the cluster for production-representative numbers.
- Setup Guide - Setup environment to run e2e tests The full HyperFleet stack (API, Sentinel, adapters, broker) must be deployed and healthy.
- Verify Configuration - Configure test settings
Ensure the environment variables are set correctly in
env/env.localand they match your infra deployment settings.
Run from the repository root:
source env/env.localFor realistic baselines, seed the database before running tests. Seeded rows add
realistic table size so query planner behavior and index performance reflect
production conditions. Clusters and NodePools share the resources table with
Channel, Version, and WifConfig (see HYPERFLEET-1159),
so seeding any of these kinds grows the same underlying table.
curl -f -X GET ${HYPERFLEET_API_URL}/api/hyperfleet/v1/clusters/
# Expected result: HTTP 200 OK.# Seed 1000 clusters
./perf/seed-clusters.sh 1000
# Check what's in the database
./perf/seed-clusters.sh status
# Clean up seeded clusters only
./perf/seed-clusters.sh cleanup
# Or delete ALL clusters (clean slate)
./perf/seed-clusters.sh resetFor the generic (non-reconciling) resource kinds — Channel, WifConfig, Version —
use seed-resources.sh, which takes the kind as its first argument. Versions are
nested under channels, so seeding versions creates (or reuses) a single dedicated
parent channel and seeds all versions under it:
# Seed 1000 channels / wifconfigs / versions
./perf/seed-resources.sh channels 1000
./perf/seed-resources.sh wifconfigs 1000
./perf/seed-resources.sh versions 1000
# Check what's in the database
./perf/seed-resources.sh channels status
# Clean up seeded resources only
./perf/seed-resources.sh channels cleanup
# Or delete ALL resources of that kind (clean slate)
./perf/seed-resources.sh channels resetMake sure you have seeded the database first for realistic baselines.
./perf/run-in-cluster.shThis builds the image, pushes it, and runs the tests in-cluster.
./perf/parse-report.shThis extracts [PERF] and [FAIL] lines from the latest output file and generates a summary report.