Nuxeo nodes and cluster setup is provisioned using ansible script
The bench execution:
- init (using ansible):
- start required services: start db, Redis, ES, Graphite
- clean services flushing data
- create nuxeo cluster nodes
- run the bench
- run gatling simulation to create users, setup ws
- run platform importer to create 500k docs, 2x 250k in parrallel
- run gatling test to set random ACLs (TODO)
- run gatling existing simulations REST import/read JSF read
- run full ES reindex
- get results (using ansible):
- get all logs and monitoring
- stop nodes and services
- generate a bench report using a static site builder (like hugo)
- deploy the reports somewhere
–
The main job to run the bench is here:
http://qa.nuxeo.org/jenkins/job/nuxeo-reference-bench/
It triggers another job to add the artifacts to the site:
http://qa.nuxeo.org/jenkins/job/nuxeo-reference-site-add/
Then the site is update with a third job:
http://qa.nuxeo.org/jenkins/job/nuxeo-reference-site/
And available here:
http://qa.nuxeo.org/benchmarks/