README
¶
Metal Pod Reaper
Automatically recover Stateful application Pods when a Kubernetes node (on metal) can safely be established to be down.
Details
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.
Safe Node Checks
- there is a flat network (single host Network)
- all node peers can detect the node is uncontactable (ping)
Usage
TBD
Build
Binaries are created in ./bin/.
To install dependencies and build:
make
To Build with dependencies and test:
make test
To build quickly:
make build
Roadmap
Metal Pod Reaper releases are detailed in the milestone page.
Documentation
¶
There is no documentation for this package.
Source Files
Directories
¶
| Path | Synopsis |
|---|---|
|
pkg
|
|
|
detector
Package detector detects if nodes are not contactable Both the source ip and the destination IP are reported
|
Package detector detects if nodes are not contactable Both the source ip and the destination IP are reported |
|
monitor
Package to detect a Quorum and and to report or invoke the reaper
|
Package to detect a Quorum and and to report or invoke the reaper |
|
mpodr
Package mpodr detects if a node is uncontactable!
|
Package mpodr detects if a node is uncontactable! |
|
reaper
Package reaper will detect if a node is uncontactable!
|
Package reaper will detect if a node is uncontactable! |
Click to show internal directories.
Click to hide internal directories.