Configuration

The Philterd Policy Editor can be configured using environment variables:

Environment Variable	Description	Default
`HIDE_PII_WARNING`	Set to `1` to hide the PII warning banner.	`0`
`CUSTOM_HEADER_FILE`	Path to a file containing custom HTML to be inserted into the `<head>` of the page.	(empty)
`CUSTOM_FOOTER_FILE`	Path to a file containing custom HTML to be inserted into the footer of the page.	(empty)

Policy schema version

The editor authors a single redaction policy schema version: the one the bundled Phileas runtime can run, so any policy built here can also be tested here. The version is shown on the page next to the application version and commit, and there is nothing to configure. Moving to a newer schema means upgrading the editor to a build whose Phileas, PhiSQL, and bundled schema all target it.

Health and metrics

Three endpoints are exposed for monitoring:

Endpoint	Description
`/api/health`	Returns HTTP 200 and the status and application version while the application is serving.
`/actuator/health`	Returns HTTP 200 with a body containing `"status":"UP"` while the application is serving.
`/actuator/prometheus`	JVM, process, Tomcat session, and HTTP request metrics in Prometheus text format.

All three are reachable without authentication.

/api/health is the health endpoint shared across Philterd products, and is the one that reports which build is running:

{"status": "UP", "applicationVersion": "1.1.0"}

The version is the one recorded at build time, so it is what is actually deployed rather than a value that has to be kept in step by hand. A build made outside Maven, such as one run from an IDE, records no version and the endpoint reports "unknown".

The two actuator endpoints match Philter. Point a monitoring system at either health endpoint and a Prometheus scrape job at /actuator/prometheus:

scrape_configs:
  - job_name: policy-editor
    metrics_path: /actuator/prometheus
    static_configs:
      - targets: ["localhost:8080"]

The editor registers no metrics of its own, so what is exported is the standard Spring Boot set: heap and garbage collection, thread and class counts, Tomcat session counts, process uptime and CPU, disk space, and http_server_requests_seconds timings broken down by URI and status.

Kubernetes-style probes are available as sub-paths of the actuator health endpoint: /actuator/health/liveness and /actuator/health/readiness. Apart from those, no other actuator endpoint responds. Two application properties control what is published:

Property	Value	Description
`management.endpoints.web.exposure.include`	`health,prometheus`	The only endpoints published. Everything else returns 404.
`management.endpoints.web.discovery.enabled`	`false`	Turns off the discovery index at `/actuator`, which would otherwise list the published endpoints.

Either can be overridden at runtime with the matching environment variable, for example MANAGEMENT_ENDPOINTS_WEB_EXPOSURE_INCLUDE.

The docker-compose.yml in the repository uses /actuator/health as the container healthcheck, so docker compose ps reports the service healthy only once it answers with an UP status. The runtime image contains no curl or wget, so the probe speaks HTTP over bash's /dev/tcp.