Dashboard Guide
A tour of the dashboard's main pages. Every page shares the same header (facility, search, local clock, light/dark theme toggle, notifications, the page tabs with the open-incident count, telemetry freshness and system count) and the same facility-scoped data. The dashboard opens in the dark theme; the toggle switches colours only and is remembered per browser.
Your first daily check
- Confirm the organization/facility and last update time. Delayed, stale, or missing telemetry is not a healthy reading.
- Read Fleet’s verdict and the Needs action row, then open an affected storage system from its tile. Compare playback, drive health, host network, capacity, and maintenance evidence.
- Open the related incident to review its evidence and timeline. Assign responsibility through acknowledgment, add investigation notes, and verify recovery after any action.
The sandbox uses simulated data. Its scenarios demonstrate workflows, not your facility’s health or measured product performance.
Your first connected system
A connection confirms that your agent is reporting. Workload readiness stays unconfirmed until the required assessments have usable evidence; healthy drives alone do not establish playback readiness. Known problems still take priority.
Review Playback & Workloads in Settings and keep the agent connected. Normal storage activity supplies latency samples, while capacity forecasts show their recorded history progress. Choose a recognizable appliance display name during setup; its technical hostname remains available in details.
1.Main Pages
| Page | Route | Purpose |
|---|---|---|
| Fleet | /dashboard | The facility at a glance, readable across a room: a verdict with counts by status, when operating hours next start and whether you are ready, how things changed since you last closed, then every storage system as a tile, worst first, in Needs action / Watch / OK rows. Wall mode (?wall=1) shows it full screen for a wallboard. A facility with no nodes yet shows the pairing CTA. |
| Storage | /storage | Every storage system in one table, worst first: status, suite and platform, a plain sentence saying why, Playback, Drives, Capacity and Maintenance as icon + word, and last seen. Filter by name, platform, health, Observer presence, capacity risk, and open incidents. |
| Drives | /drives | Which drives need attention, and where they sit. Systems with a flagged drive are drawn bay by bay with a worded callout per drive (Replace now, Replace soon, Watch, Check drive); every other system is a compact card with its bay shape and a status line. Bays come only from the enclosure mapping; systems without one list drives by device name. Every bay links to that drive's detail. |
| Incidents | /incidents | Full incident list across open / acknowledged / resolved / all, filterable by severity, node, diagnostic type, and time window. |
| Maintenance | /maintenance | Reported scrub/resilver/parity activity, projected completions, and operating-hour collisions. Future host schedules are not collected; completed history depends on reported timestamps. |
| Capacity | /capacity | Every pool ranked by shortest runway to its capacity warning threshold, with 7-day and 14-day at-risk counts surfaced up front. |
| Chargeback | /chargeback | Dataset-to-project allocation and statement exports on eligible plans. All roles can view; Billing Admins and Organization Admins can manage allocations. |
| Security (posture) | /security-posture | Technical deployment-control evaluation with per-node evidence, history, and role-controlled exports. It is not an external compliance certification. |
| Settings | /settings | The eight configuration sections — see the dedicated Settings reference. |
Reading the Fleet wall
- Status. Critical (red): an impacted system, a critical incident, or a faulted pool. Warning (yellow): any other open problem, playback near its limit, a maintenance collision, or a forecast risk. Offline or No data (grey): Pillar cannot vouch for the system. OK (green): nothing to report. Every status is an icon and a word.
- Tiles. Each problem tile shows how long the worst problem has been open, the system, a few words on why, and its suite (or platform when unmapped). Click a tile to open the system.
- Tomorrow and Since. Built from your operating hours: whether the facility is ready for the next opening, and how many problems opened or resolved since it last closed.
- Large facilities. Past 24 systems, warnings become compact rows and OK systems small chips grouped by area; past 120, each area is one tile and areas where everything is OK fold to one line.
- Wall mode. The Wall mode button opens a full-screen view for a wallboard: no tabs, large type, the facility clock, and Exit wall mode. It refreshes every 60 seconds; Pause live updates holds it still.
2.Per-Node Investigation
Clicking any storage system opens a node investigation view with tabbed sections:
overview, playback, network, capacity, maintenance, topology, drives, incidents — reachable as a full page (/storage/<node>?tab=<tab>), which Fleet tiles and Storage rows open, or as an inline drawer on the Fleet page (/dashboard?node=<node>) used by links from search, notifications, and incidents.
Overview answers whether the system is safe for playback: a verdict in words with the likely cause, the facility's operating hours, the playback streams affected and the first next step; a health checklist (Playback, Drives, Pool, Capacity, Maintenance, Network), each row an icon, a word and one sentence linking to the tab with the evidence; a playback headroom table by frame rate (frame budget, time left after a read, verdict); one read-latency chart for the last 6 hours against the safe target; what Pillar can observe on the host; and the open incidents.
3.Search
Global search (facility-scoped) covers nodes, pools, datasets, drives, enclosures, incidents, and — where configured — production suites and projects, so an engineer can jump directly to the object they're investigating instead of navigating through each page.
4.Notifications
The header bell shows recent notifications with a dedicated Notifications page for the full history (up to 200 most recent), with per-user seen state. Marking a notification seen does not acknowledge or resolve its incident, and does not prove external alert delivery.
See also