Grokking DevOps and SRE Interviews
0% completed
Upgrading and Operating a Cluster
These failures have all followed a routine Kubernetes upgrade:
- A Helm chart, a package of manifests (the YAML files that describe Kubernetes objects), still uses an API version that the new release no longer serves, so the next deployment fails.
- A node drain waits for hours, because a PodDisruptionBudget allows zero evictions.
- A new node starts from an old operating system image that uses cgroup v1, and its kubelet refuses to start.
- The control plane fails during the upgrade, and the newest backup of its data is a month old.
None of these comes from a bug in Kubernetes
.....
.....
.....
Like the course? Get enrolled and start learning!
Reading Progress
0%