Make the PDS Your Own: Customization and Metrics

The release consolidates all UI branding options into a documented set of environment variables, adding support for a background image on authentication screens. Operators can apply these settings by editing pds.env and restarting the Docker container, which is the only required step. This approach works for the reference PDS deployment; custom deployments that bypass the provided Docker compose would need to replicate the variable handling in their own startup scripts. No code changes are needed for the UI itself, but the variables are ignored if the PDS is not built from the reference source.

Observability is introduced through optional OpenTelemetry instrumentation. By setting NODE_OPTIONS and OTEL_* variables, the PDS begins exporting traces, metrics, and logs to any OTLP endpoint, with automatic coverage of HTTP traffic, SQLite, S3-compatible storage, and Node runtime metrics. The added counters for account creation, session creation, and OAuth authorizations give operators concrete data points for monitoring user activity. Telemetry is disabled by default, so enabling it incurs only the cost of the environment configuration and any downstream collector resources; the reference monitoring stack adds roughly 1 GB of memory usage if used as provided.

The default maximum blob size has been raised to 300 MiB, which directly impacts the size of files that can be uploaded without explicit configuration changes. This change benefits services that store longer video clips or high-resolution media, but it also means that storage and bandwidth consumption may increase proportionally. The limit remains a ceiling; actual upload size is still constrained by the underlying storage backend and any external quotas, so operators must ensure their storage layer can handle the larger payloads.

A bug affecting the connected apps list has been fixed, requiring no action from operators beyond upgrading to the new release. The fix is limited to the reference implementation and does not affect other components of the system. As with the other changes, the fix is delivered via the updated Docker image.

A ready-made monitoring stack is included in the repository under the monitoring/ directory, bundling Prometheus, Grafana, and node_exporter with a pre-built dashboard that visualizes host health and PDS activity. The stack can be run alongside a PDS instance using the provided compose file, or operators can import the dashboard JSON into an existing monitoring setup. By default all services bind to localhost, so external access requires an SSH tunnel or explicit reconfiguration of the OTEL endpoint.