Building a zero downtime Kubernetes cluster
summary
Catalyst Cloud builds and runs an open source Magnum Auto Healer for OpenStack Magnum that detects unhealthy masters, etcd or worker nodes via API endpoints and node conditions and repairs them one-by-one, preserving original counts (e.g. three masters stay three). It uses leader election, a configurable monitor interval (default 15m) and a check delay to avoid jitter, integrates with cluster-autoscaler (workload and min/max node tuning required), runs in production and restores a demo node in roughly five minutes.
AI-written summary · about
chapters (15)
- 0:00Introduction
- 1:37About Feilong Cloud
- 3:37Why are you interested in communities
- 4:45Zero down time
- 6:57Worry little tricks
- 8:58Additional tips
- 10:35Communities
- 13:07Maintain
- 15:19Unexpected failure
- 17:26Autohealing
- 27:35Magnum Auto Healer
- 30:49Config File
- 33:49Scaling
- 37:43Autofilter
- 40:19Questions
related talks
34mEffortless On-Premise Kubernetes: Cloud-Native Experience at a Fraction of the Cost
30mMission Critical Kubernetes
40mOperating a Self-Healing Bare-Metal Kubernetes Platform at Global Scale
28mThe Magic of Kubernetes Self-Healing Capabilities
30mEdge Kubernetes and Healthcare: A Next-Gen Architecture that Scales Across Thousands of Locations
35mAcing Distributed Container Management
32mAutomating and Testing Production Ready Kubernetes Clusters in the Public Cloud
18mAutomating Kubernetes Cluster Updates: Achieving Zero Downtime Effortlessly