Record the healthy path
A traceroute made during an outage is only half a story. Keeping small baseline measurements from healthy periods gives later packet captures and route changes something useful to compare against.
Notebook
Operational notes
Make configuration changes boring
Save the effective configuration, make one change, validate it, then keep a rollback path. This is less exciting than emergency repair and considerably more pleasant.
Monitor certificates as dependencies
A running process with an expired certificate is not a healthy service. TLS renewal and public-path checks belong next to ordinary process monitoring.