The dashboard

/ (Overview) is the landing page after sign-in, kept short on purpose:

• Today's numbers under the greeting, as four tiles: how many services are running out of how many, and today's requests, average response time and error rate, each with how it moved against yesterday.
• Traffic, today's requests in two-hour bars, the busiest one highlighted with a dashed line at its height. Today opens Monitoring.
• Needs you, everything waiting on you, the first four shown (See all for the rest): setup issues (Traefik's configuration re-applied in one click when that's the fix), the server past a hard resource limit, failed update trials (retry in one click), services whose uptime probe fails, the unresolved error issues and error-level log lines of the last 24 hours, and the services with an update waiting (applied in one click). It updates live, and says so when nothing needs you.
• Deployments, the in-flight and latest deploys, pushed from the server as they move: the service, what it shipped (commit or image tag), when it started, how long it took and its status. Needs read access to Services.
• Host, live CPU, memory, disk and (with an NVIDIA card and nvidia-smi) GPU for the machine Homerun runs on, refreshed every five seconds.
• Deploy an app, in simple mode only: popular templates with a Quick Deploy button each.

Their history, over a day to all time, is on Monitoring.

Every chart reads out what's under the pointer: a bar's time range, requests and error share, a line's value at that moment with its unit, a ring's or meter's used and total, an uptime beat's time, state and latency. Tap a chart on a phone, or focus it and use the arrow keys, for the same reading.

• Resource limits: each of CPU, memory, disk and GPU has a soft and a hard threshold (Settings → General → Resource limits, 85% and 95% by default), checked against the host once a minute. A resource only alerts once it has stayed past a limit for a while (Alert after, 60 seconds by default, so a short spike doesn't page you; the disk alerts at once, it doesn't spike), which opens an incident: past the soft limit it sends a Resource warning, past the hard one (or when it climbs there later) a Resource critical, shows a red banner here, and refuses new services (from the wizard, a template, a compose import or migration, the API, or a pull request preview) until usage drops back. While the incident lasts it reminds you (Remind every, 30 minutes by default, 0 for never), and the first reading back under the soft limit sends a back to normal for that resource with how long it lasted and its peak. The last incidents are listed under the limits in Settings, and one still open when Homerun restarts stays open: no second alert, and you still get its recovery. If the host's stats can't be read, nothing is refused.

#List pages

Every list page (services, stacks, templates, cron jobs, storage, remote hosts, build cache registries, backup destinations, users) has the same toolbar: a search box, a sort (newest, oldest, name A–Z or Z–A, recently updated, and most services on stacks), filters where the list has any, and a pager. All of them run on the server over everything you can see, and live in the URL.

#Search

The Search… button in the header, or ⌘K / Ctrl+K anywhere, opens a command palette. It jumps to any dashboard page by name, and once you've typed two characters it also searches the instance's services, stacks, templates, cron jobs, storage volumes, backup destinations, remote hosts, build cache registries, git providers, notification channels and status pages, plus users and authentication providers for an admin.

#Setup diagnostics

Homerun runs a handful of read-only checks on every dashboard load: base domain and auth secret still at their defaults, no Dashboard URL set, the dashboard's own hostname actually routed to it by Traefik, the Traefik container reachable, the worker reachable and the Docker socket it fronts answering, Traefik still running with the flags your settings put on it (the swarm provider, the HTTP cache plugin, the ACME email), Docker not stuck on a helper container (see Docker stuck on a container), and SMTP fully configured if you turned it on. The worker and Docker checks are reported separately on purpose: a worker that's down and a worker that's up but can't reach Docker are different problems with different fixes. Any that aren't OK show up under Needs you on the dashboard.

Each one links straight into /settings, on whichever tab the offending field lives, and rings the relevant fields so you can see what it's complaining about rather than hunting for it. Two checks have nothing to deep-link to: the auth secret is env-only (AUTH_SECRET, change it and restart), and the Traefik checks are about a live container rather than a setting. When Traefik has lost flags your settings put on it, usually because it was recreated from the compose file, an admin gets a Re-apply button that puts them back (Traefik restarts, so routes drop for a few seconds). Homerun also re-applies them on its own every time its worker starts.

There's no separate /setup page, Needs you plus /settings is the whole flow.