Health Checks

Health Check probes your origin (backend) servers automatically at regular intervals. An unhealthy server is removed from rotation; when it recovers, it is automatically re-added — visitors never hit a server that is down.

Note

Health checks are opt-in and configured per origin. If you don’t enable them, HEIMWALL distributes traffic across all configured servers without monitoring their status.

---

Enabling and configuring

1. Go to Console → WAF → [Your domain] → Origins.

2. Click the edit icon on the server row you want to configure.

3. Turn on the Health Check toggle. If this is the first time, read the firewall warning and check the confirmation box (see below).

4. Fill in the basic fields:

   Field	Description	Default
   Method	HTTP method (GET / HEAD)	GET
   Path	URL path to probe	/
   Expected status codes	HTTP response codes to treat as healthy, comma-separated	2xx/3xx + 401/403

5. Expand Advanced settings if needed:

   Field	Description
   Interval (s)	Time between probes
   Timeout (s)	How long to wait for a response (must be at most half the interval)
   Failure threshold	How many consecutive failed probes before the server is disabled

6. Click Save.

Add HEIMWALL IP blocks to your firewall

Probes come from variable IP addresses across the HEIMWALL edge network — there is no single fixed address. If your origin server’s firewall restricts inbound connections, probes will be blocked and a perfectly healthy server will be marked unreachable.

Before enabling: add all IPv4 and IPv6 blocks published at heimwall.com/ips to your origin firewall allowlist.

---

How it works

At each interval, a probe is sent to every origin server. If the server responds within the timeout with one of the expected status codes, it is counted healthy. Otherwise — timeout or unexpected status code — the probe is counted as a failure.

When consecutive failures reach the failure threshold, the server’s status is set to passive and no traffic is sent to it. When it starts responding healthily again, Health Check automatically re-enables it.

Important: Health Check only re-enables servers that it disabled. It does not touch servers manually disabled by an administrator.

Why timeout can’t exceed half the interval: if timeout were greater than half the interval, probes would overlap. A slow-but-working server could be marked unhealthy before it gets a chance to respond. For example, if the interval is 10 s, the timeout can be at most 5 s.

---

Failover behaviour

• Failover happens at the WAF layer — the DNS record does not change and visitors notice nothing.
• Load Balancer on: the passive server’s load is redistributed to remaining healthy servers.
• Load Balancer off: the next healthy server in the list takes over.
• Works alongside Connection Failover: Health Check proactively removes servers from the pool; Connection Failover handles connection errors that arrive before a probe has had time to mark the server unhealthy.

---

Monitoring

From Console → WAF → [Your domain] → Health Check you can track:

Metric	Description
Uptime %	Ratio of successful probes in the selected time range
Average RTT	Average probe response time and range
Events table	Down/up transitions with time, server, reason and who detected it

Time range selector: 1 h / 6 h / 24 h / 3 d / 7 d.

The events table also shows the source of each status change: Health Check (automatic) or Manual. This lets you immediately tell whether a server went passive on its own or was disabled by an administrator.