Adding our own scripts to https://github.com/kubernetes/node-problem-detector and sharing them in case you might find those handy for your use cases.

The script details can be found in /config/plugin/ but ultimately, they are:

• launch-config-drift: a way to check if your instances launch template has diverged from your ASG launch template
• spot-termination: uses the meta-data/spot/instance-action endpoint to check EC2 Spot Instance interruption notice
• local-dns-resolver: checks the response status value received (if any) from the local DNS resolver IP
• upstream-dns-resolver: check if we receive an IPv4 address for a given A record
• uptime: every 5 seconds, checks if the information detailing how long the system has been on since its last restart is acceptable (to us the threshold being 604800 seconds)

July 2024 - The custom node problem detector image is now stored in the uswitch/node-problem-detector repository on Quay.

Base image tags available at: https://explore.ggcr.dev/?repo=registry.k8s.io/node-problem-detector/node-problem-detector

To publish a new tagged image to Quay, push a semantic version tag to the repository:

# Ensure you're on the latest master
git checkout master
git pull origin master

# Create and push a tag
git tag v1.0.0
git push origin v1.0.0

This triggers the CI workflow which builds and pushes the image to quay.io/uswitch/node-problem-detector with both the version tag and the commit SHA.