Self-Heal Recovery
Automated service recovery that restarts crashed mail services and alerts you only when automatic recovery fails.
Self-Heal automatically detects and restarts crashed mail services without manual intervention. If a service cannot be recovered, an email alert is sent to the admin so they can investigate.
Business feature
Self-Heal Recovery is available on the Business plan and above. The Pro plan includes service watchdog with email alerts (without automatic recovery).
How it works
The health watchdog runs every 30 seconds. On Business and above, when a monitored service is detected as down:
- Automatic restart -- the watchdog attempts to restart the service immediately
- Verification -- waits a few seconds and checks if the service is running again
- Success -- if the service recovers, no alert is sent. The dashboard banner clears automatically
- Failure -- if the service cannot be restarted, an email alert is sent to the admin
This means most transient crashes (memory pressure, assertion failures, brief lock conflicts) are resolved automatically within minutes, without the admin ever needing to intervene.
Covered services
Self-heal monitors and recovers all mail server services:
| Service | Role |
|---|---|
| Postfix | Mail transfer agent (SMTP) |
| Dovecot | Mailbox access (IMAP/POP3) |
| OpenDKIM | DKIM email signing |
| MariaDB | Database for domains, users, aliases |
| SpamAssassin | Spam filtering |
| SpamAssassin Milter | Milter bridge connecting SpamAssassin to Postfix |
| Nginx / Apache | Web server (reverse proxy for dashboard and webmail) |
| Unbound | Local DNS resolver |
| Rsyslog | System logging |
| Fail2ban | Brute-force protection |
| CeyMail AI Filter | AI-powered email screening (if enabled) |
When self-heal cannot recover
Some failures require manual investigation:
- Configuration errors -- a corrupted config file will cause the service to crash again immediately after restart
- Disk full -- services cannot start if the disk is full
- Port conflicts -- another process may be using the required port
- Corrupted database -- MariaDB may need manual recovery if tables are corrupted
In these cases, the self-heal attempt fails, the email alert is sent, and the dashboard banner remains visible with a Restart button for manual intervention.
Comparison with Pro watchdog
| Feature | Pro | Business |
|---|---|---|
| Dashboard banner for all services | Yes | Yes |
| Email alerts for mail-critical services | Yes | Yes |
| Automatic restart on failure | No | Yes |
| Alert only if restart fails | No | Yes |