PerfPulse

In-cluster k6 workload runner for CANFAR performance evidence. See CONTEXT.md for domain language and docs/ for runbooks and ADRs.

Local development: bun run check.

Skaha workload image (stress-ng)

Skaha sessions use images.canfar.net/skaha/stress-ng:latest (Ubuntu + stress-ng only). The image is rebuilt manually when needed; charts reference the :latest tag.

Dockerfile: docker/stress-ng/Dockerfile

From the repo root:

docker login images.canfar.net

docker build \
  --platform linux/amd64,linux/arm64 \
  --tag images.canfar.net/skaha/stress-ng:latest \
  -f docker/stress-ng/Dockerfile \
  .

docker push images.canfar.net/skaha/stress-ng:latest

Quick check under the fixed workload footprint (1 CPU, 1 GiB):

docker run --rm --platform linux/amd64 --cpus 1 --memory 1g --memory-swap 1g \
  images.canfar.net/skaha/stress-ng:latest \
  --cpu 1 --temp-path /tmp --timeout 5s --metrics-brief