- ReportServer and template editor: metrics for Prometheus and Grafana at `GET /metrics` (text format 0.0.4, no new dependency; a small shared registry in `NetReporter.ReportServer.Client`). The endpoint needs no login but answers only on its own port (`Metrics:Port`, `METRICS_PORT`, default 9464, `0` switches it off); the decision is made by the local port of the connection, not by the `Host` header, so it is `404` on the public port 8080 even for an admin. The Docker images of the API and the editor listen on 8080 and 9464 (`ASPNETCORE_URLS`); requests to `/metrics` on the API are logged at Debug. - API metrics: renders (`netreporter_renders_total{format,source,outcome}` with outcome success, cache_hit or failure, `netreporter_render_failures_total{format,code}`, the histogram `netreporter_render_duration_seconds`, pages and output bytes), requests refused by a rate limit (`netreporter_rate_limited_total{policy}`), password logins (`netreporter_logins_total{result}`), the content cache (`netreporter_content_cache_*`: state, hits, misses, bytes served from disk and bytes read from the database) and an inventory read from the database at most once a minute per pod (`netreporter_content_items{kind,visibility,origin}`, published templates and versions, assets by kind and size, users and users logged in during the last 24 hours and 30 days). - Editor metrics: open editor sessions (`netreporter_editor_sessions`), previews (`netreporter_editor_previews_total{format,outcome}` and the duration histogram), gallery thumbnails, template saves (`created`, `updated`, `conflict`) and publishes (`published`, `unchanged`, `conflict`). Both processes also report memory, the number of compiled C# expressions with its cap, uptime and the version (`netreporter_build_info`). - k3s: `deploy/k8s/05-workloads.yaml` names the container port `metrics` of the API and the editor; the new `deploy/k8s/09-monitoring.yaml` holds a `PodMonitor` for kube-prometheus-stack and two Grafana dashboards (cache and process of the API; renders, content, users and editor work). `rollout.sh` and `deploy.sh` apply it only when the cluster has the `PodMonitor` CRD. The dashboards are also plain JSON files in `deploy/grafana/` that can be imported into any Grafana; they are the source, and `deploy/grafana/sync-k8s-dashboards.ps1` copies them into the ConfigMaps. Celý seznam změn: CHANGELOG.md v balíčku a https://dev.azure.com/advsoft/Knihovny/_git/NetReporter?path=/src/NetReporter/CHANGELOG.md