Drain-Node-On-Crash

This app is designed to automatically Drain a node after a crash where the node fails to recover after 5mins.

Install

git clone https://github.com/mattmattox/drain-node-on-crash.git
cd drain-node-on-crash
kubectl apply -f .

Upgrade

rm -rf drain-node-on-crash
git clone https://github.com/mattmattox/drain-node-on-crash.git
cd drain-node-on-crash
kubectl apply -f .

Default settings

NODE_TIMEOUT = 360 (seconds)

AUTO_UNCORDON = true (This setting will automatically uncordon a node that was drained by the script. NOTE: Nodes that have cordon outside this app will not be changed.)

REMOVE_PODS = true (This setting will automatically delete all pods from the node after draining.)

CATTLE_CLUSTER_AGENT = true (This setting will apply a fix for cattle-cluster-agent.)