9 checks after every Linux reboot

Quick reference

# Nothing failed
$systemctl --failed
# Booted the right kernel
$uname -r
# Every disk is mounted
$df -h

# IP and route look right
$ip -br a; ip r
# Services are listening
$ss -tlnp
# Clock is in sync
$timedatectl

# No new errors
$journalctl -p err -b
# The app answers
$curl -sI localhost
# Scheduled jobs still there
$systemctl list-timers

systemctl --failed — any unit that did not start. An empty list is what you want to see.
uname -r — the kernel that is actually running. After a kernel update, this confirms the new one booted, not the old one.
df -h — every filesystem you expect is mounted. A data disk that did not mount means your app may be writing into an empty folder on the root disk.

ip -br a; ip r — the right IP addresses on the right interfaces, and a default route. Short output, easy to read.
ss -tlnp — the services that should be listening are listening, on the ports you expect.
timedatectl — "System clock synchronized: yes". A wrong clock breaks TLS, logs and anything with tokens.

journalctl -p err -b — errors since this boot only. New errors after a reboot deserve a look before anyone reports them.
curl -sI localhost — the app actually answers, not just the process running. Use your real health endpoint if you have one.
systemctl list-timers — backups and scheduled jobs are still there and have a next run time.

This takes about five minutes. The alternative is finding out from a user, or from a backup that silently did not run.

Reboot is not the last step. Checking is.

Reboot is not the last step. Checking is. Five minutes now saves a long night later.

A one-line first look

$systemctl --failed; df -h; journalctl -p err -b