Monitoring and alerts
Health checks, uptime, certificates, logs and alerts by email, Slack or webhook.
airepeat shows only readings that have a real source. When something can't be measured, it says "Unavailable" and why, rather than showing a number.
Health checks and uptimeLink to this section: Health checks and uptime
Every 5 minutes airepeat checks each launched site's health address (/api/health). Results are kept for 7 days.
- Uptime is the share of passed checks in the last 24 hours.
- Response time is the median of the last day's successful checks, timed from airepeat, so it includes the network.
- Three failures in a row mark the site degraded, and the next success marks it live again.
A degraded site offers Ask AI to look, which starts one task with the recent checks. It fixes the cause if it's in the code, and changes nothing and says so when it isn't.
Server readingsLink to this section: Server readings
CPU, disk and network activity for servers come from Hetzner. Hetzner doesn't report memory or how full a disk is, so those show as unavailable. Coolify doesn't report a single app's CPU or memory through its API either.
LogsLink to this section: Logs
The site's Logs tab follows the app's output live. Credentials, API keys, tokens and the local part of email addresses are masked. Build logs are kept for every deploy.
CertificatesLink to this section: Certificates
Each launched site's TLS certificate is read at most every 12 hours. Monitoring shows its expiry and issuer, and alert rules can use the days left.
AlertsLink to this section: Alerts
Create alert rules on uptime, certificates and other metrics. An alert whose metric can't be read is unknown, never resolved. Firing alerts notify at most once per cooldown, by:
- email, through the SMTP account your workspace connects;
- Slack;
- a webhook, signed with HMAC-SHA256.
Deliveries are tried up to three times.