Uptime monitor and alerts for the cloud server

The cloud server checks itself every five minutes and tells you when something needs a person. You will be able to choose where the alerts go.

Before you start #

• You need the server's administrator access (the person who runs provision.sh).
• Pick where alerts go: an e-mail address, a web address from your chat tool, or both. Both are free to set up.

Steps #

1. Open /etc/psd/monitor.conf and add the lines you need:
   • PSD_ALERT_EMAIL=you@example.com
   • PSD_ALERT_WEBHOOK= followed by the incoming-message address from your chat tool
   • Optional: PSD_ALERT_AFTER=3 (failures in a row before an alert), PSD_DISK_MIN_FREE_PCT=10, PSD_BACKUP_MAX_HOURS=30, PSD_CERT_MIN_DAYS=14.
2. Run /opt/psd/cloud/monitor.sh --test. A test alert arrives on every channel you set.
3. The monitor runs every five minutes from then on. provision.sh switches the timer on.

What it checks #

• The health address of the main host and of every business listed in tenants.txt.
• Free space on the data disk.
• The age of the newest backup.
• Days left on the web certificate.

Each problem sends one alert after it has failed three times in a row, so a short blip stays quiet. When it clears you get one "Recovered" message.

If something goes wrong #

• No test alert: the log line says which channel failed. For e-mail, check that the server can send mail. For the web address, check the address.
• An alert that repeats every run: the send failed, so the monitor tries again on the next run.