Automatically recover Stateful application Pods when a Kubernetes node (on metal) can safely be established to be down.
In Kubernetes, when there is no Cloud Provider integration, there is no automated way of automatically recovering workloads (pods) when the status of a node is Unknown / NotReady.
- there is a flat network (single host Network)
- all node peers can detect the node is uncontactable (ping)
TBD
Binaries are created in ./bin/.
To install dependencies and build:
make
To Build with dependencies and test:
make test
To build quickly:
make build
Metal Pod Reaper releases are detailed in the milestone page.