Failover from OpenStack to AWS in BaseOS CI/TFT
summary
Testing Farm builds a standalone service called Artemis that transparently fails over VM provisioning from OpenStack to AWS (and can fall back to Beaker) using a Python programmable routing script, drivers for each cloud, a queue+broker+DB backend, REST API and a JSON CLI. Artemis retries transient API errors, tracks pool metrics in Grafana and Sentry, and supports cloud bursting; it runs in production for about six months and already prevents irrecoverable CI failures and unexpected quota blips.
AI-written summary · about
chapters (10)
- 0:00Introduction
- 0:31Agenda
- 2:04Previous Testing Pipeline
- 4:06Artemis
- 5:48Artemis Internals
- 8:05Pools
- 10:15Routing
- 12:34Metrics
- 15:50Artemis CLI
- 16:49Inspect Request
related talks
01h 10mHosting Drupal on AWS
34mInfrastructure at Scale Using ArgoCD
17mDeployments We Can Believe In
01h 08mPerformance testing in the cloud
29mLessons Learnt Operating Multiple Cloud Foundry on OpenStack Deployments
51mBuilding a multi-account, multi-runtime service-oriented architecture
26mTowards a Fault-Resilient Cloud Management Stack
01h 32mUsing Terraform and AWS to create environments on the fly