Monitoring
Read live metrics, set threshold and uptime alerts, and route them to channels.
Monitoring shows how your services are doing and tells you when something breaks. It covers metrics, alerts, notification channels, and log drains.
Metrics#
A service’s Metrics tab charts requests, errors (4xx and 5xx), latency (p50 and p95), bandwidth, CPU, and memory. Choose a window of 1 hour, 24 hours, or 7 days. The charts refresh every 30 seconds.
Metrics come from Prometheus. If the metrics stack isn’t configured, the tab explains how to enable it.
Alerts#
In settings, admins can add alert rules that fire when a metric crosses a threshold:
- Metric
- CPU (cores), Memory (MB), or Service down.
- Condition
- Greater than or Less than a threshold you set.
- Scope
- A single service, or all services.
Notification channels#
Route events to an email address or a webhook URL. Subscribe to specific events, such as deploy succeeded, deploy failed, member joined, quota warning, or SSL expiring, or leave them all unchecked to receive everything. Use "Send test" to check delivery. Set a signing secret to add an X-Gheima-Signature HMAC header.
Log drains#
Add a log drain to forward deployment logs to an external HTTP endpoint (for example your logging provider). Logs are POSTed as they finish, optionally signed.
History & delivery#
- Each alert keeps a history of its state changes between OK and Firing.
- Channels record every delivery with its HTTP status; failed ones can be retried.
- The metrics API takes a window of 5 to 360 minutes. The Metrics tab offers 1 hour, 24 hours, and 7 days.
- Signed deliveries include an X-Gheima-Signature HMAC header you can verify.
✦ You see resource usage live and get notified by email, webhook, or your own tooling when something needs attention.