Kubernetes Perfdash

Kubernetes Perfdash (performance dashboard) is a web UI that collects and displays performance metrics. Performance metrics are created based on performance test results for different nodes numbers, platform types and platform versions.

Perfdash is available at http://perf-dash.k8s.io/

Supported metrics

• Responsiveness
• Resources
• PodStartup
• TestPhaseTimer
• RequestCount
• RequestCountByClient

Metrics above are available for all kinds of tests divided into load and density subtypes.

Application server

Application server runs as a deployment on kubernetes cluster. It is hosted on mungegithub cluster in k8s-mungegithub project.

How to deploy new version of perf-dash.k8s.io

• Increment TAG in the Makefile (for example: 2.10 -> 2.11)

• Modify the perfdash image version in the deployment.yaml to be the same as the one specified in the Makefile

• Submit a PR, get required approvals and wait until it's merged

• Run make push to push new image to container registry

• Refer to perf-dash.k8s.io/README.md for the deploy instructions

  Note: Since 2020-04-21, perf-dash.k8s.io is hosted in the public aaa (sic!) cluster. Instructions on connecting to aaa cluster can be found here.

Application images

Images are stored in gcr.io/k8s-testimages project container registry.

Local development

To test your changes locally, execute make run. It will build the binary and start perfdash website at http://localhost:8080. Note that it might take a short while for perfdash to start since it needs to fetch the job artifacts first.

You can alter startup parameters (like number of jobs for which history is fetched) by editing the makefile.