How we monitor small clusters

Operations

A short list of checks we run on every host, what we alert on, and what we deliberately ignore so the pager stays quiet.

Disk space is a feature

Infrastructure

Why logs, backups and temporary files should have limits long before the disk fills up, and how to set them.

Planning a maintenance window

Process

Pick a quiet hour, write the rollback first, and tell people what to expect. A boring change is a good change.