18.4Health endpoints
Last updated
The platform publishes health endpoints for load balancers, reverse proxies, uptime monitors and orchestration tools. They require no sign-in and return no sensitive information. The paths are relative to the address where the API is published; for example, if the API is published under /backend, the readiness probe is https://gateway.example.com/backend/health/ready.
| Endpoint | Answers | Use it for |
|---|---|---|
GET /health/live | 200 Healthy while the process is running. It checks no dependencies. | Deciding whether the process needs a restart. |
GET /health/ready | 200 Healthy when the platform can serve traffic, 200 Degraded when it can but something needs attention, or 503 Unhealthy when it cannot. | Deciding whether to send traffic to this instance, and uptime monitoring. |
The readiness probe checks two things:
- Database – the platform can reach its database.
- Audit store – the platform can record operations in its audit trail. If audit writes are failing, the instance reports itself as not ready, so that it stops receiving traffic rather than making directory changes it cannot record.
The probe answers Degraded (still 200, so the instance stays in service) when audit writes succeed but the audit trail's tamper-evidence is switched off, or when the local folder that holds audit records during a database outage cannot be written to. Configure your monitoring to alert on Degraded as well as on a failed probe.
Invoke-WebRequest -Uri "https://gateway.example.com/backend/health/ready" -UseBasicParsing |
Select-Object StatusCode, ContentTip: The update agent uses the readiness probe after installing an update to confirm that the new version started correctly. See Chapter 19, Licence, updates and backups.
Where to look when something is wrong
| Question | Where to look |
|---|---|
| Who changed this user, group or setting? | Operations log (Chapter 9) |
| Is the platform up and able to serve requests? | /health/ready |
| What error did the server just report? | Live Logs, Live view |
| What happened overnight or last week? | Live Logs, Files view, or C:\Logs on the server |
| Why is a request slow? | Telemetry traces in your collector, with Directory calls (LDAP) switched on |
| How do I keep server errors in our SIEM? | Log Stream, with a destination at Warning or Error level |
| How do I get server errors as issues in Sentry? | Log Stream, with a Sentry destination sent as Issues at Warning or Error level |
| Why does a Sentry destination say "HTTP 429"? | Sentry is rate limiting the project (quota or spike protection); check the Sentry plan. Delivery resumes by itself |
| Did an email or notification go out? | Notification history (Chapter 16) |
Last updated
Thank you for your feedback.