Providing Continuous Integration at CERN
summary
CERN runs a CI platform serving 15,000 active users and executing about 100,000 jobs per month across GitLab CI, Jenkins + Nexus, a Docker registry and distributed S3 caching. Puppet-managed runners gave way to OpenStack-backed pools (Magnum and Docker Machine), then to Swarm-based and Docker-executor runners with a split between unprivileged runners and experimental privileged runners for nested or device-needy builds. Priorities are faster provisioning, image caching, autoscaling runners and replacing special docker-build runners with an unprivileged build tool.
AI-written summary · about
chapters (14)
- 0:00Introduction
- 0:24What is Continuous Integration
- 0:59Why we need Continuous Integration
- 1:33What we do
- 2:30How we evolved
- 3:14Runners
- 3:56Puppet
- 4:37Kubernetes
- 6:11API fountains
- 8:14Dr Bit
- 10:04Other Services
- 10:16Jenkins Template
- 10:41artifact storage
- 11:06future plans
related talks
35mJenkins Scaling and Organization for an Efficient CI
29mA Multi-Cluster, Multi-Cloud Infrastructure with GitOps at CERN
36mThe trials and tribulations of a CI pipeline on Apache Infra
30mContinuous Integration; Tips & Tricks
13mEvolution of containers in continuous integration
36mCI/CD BoF: Delivering Quality Quickly
06mBuilding Continuous Delivery Pipelines
26mCNCF Cross-Cloud CI Intro