> ## Documentation Index > Fetch the complete documentation index at: https://docs.flashduty.com/llms.txt > Use this file to discover all available pages before exploring further. # RUM custom dashboards > Turn an Explorer analytics question into a Flashduty RUM custom dashboard: panel types, panel queries, dashboard variables, saving from the Explorer, table drill-down, version history and the trash. ## Overview A custom dashboard is an extra tab in the **Insights** tab bar, next to the built-in Overview, Performance, Errors and Resources tabs. Each dashboard holds panels (charts), and a panel's query is the Explorer's analytics question: event type + aggregation + group-by field + DQL filter. That makes a one-off analysis such as "which browsers do the errors come from" or "which page loads slowest" a dashboard you can keep coming back to, without rebuilding the query. ## Entry points and routes | Where | Address | Notes | | - | - | - | | Dashboard tab | `/rum/dashboard/custom/:id` | Opens one dashboard; it shows as a tab of Insights | | Edit mode | `/rum/dashboard/custom/:id?edit=1` | Opens straight into editing; changes go into a browser draft first | | New dashboard | `/rum/dashboard/custom/new?edit=1` | From the **+** menu at the right of the tab bar | | Dashboard management | `/rum/dashboard/manage` | The dashboard list: search, delete, trash | The **+** menu at the right of the tab bar has two items: * **New dashboard**: opens an empty dashboard in edit mode. This needs the **Dashboards Manage** permission; without it, the item is not shown. * **Manage dashboards**: opens the management page; viewing it needs no edit permission. The tab bar lists every dashboard by title, up to 100; beyond that, reach the rest from the management page. ## Application and time range * A dashboard follows the **application** and **time range** in the top-right corner of the page. When you open a dashboard whose link carries no time range, the dashboard's own default time range applies; changing the time range on a dashboard also changes the page's time picker. * Switching tabs carries the page's application and time range over. A dashboard's `env`, `service` and `version` variables and the matching filters of the built-in tabs are translated into each other; other variables are not carried across tabs. ## Panel types The panel editor's **Visualization** offers four kinds: | Type | Description | | - | - | | **Time series** | The measure over time, drawn as lines or bars | | **Table** | One row per group, with drill-down (see [Table drill-down](#table-drill-down)) | | **Stat** | One key number, with optional threshold colors; a stat does not use a group-by | | **Text** | Markdown text, bound to no query | ## Adding a panel Open the dashboard and click **Edit** in the top-right corner, or open a link with `?edit=1`. Edit mode is one browser draft submitted as a whole; a page refresh loses the draft, and leaving warns you first. Click **Add panel** to open the panel editor. A dashboard organizes panels into tabs and sections; a new dashboard starts with one tab. Pick the application (or **Page application**), the event type, the aggregation and measured field, the group-by field and the number of groups to keep, then write a DQL filter if you need one. The preview runs once the query is complete: after a picker change it reruns once the change settles, and a DQL filter runs on Enter or blur. Saving does not require a successful preview. A title you leave empty is generated from the query, such as "P75 Time to First Byte · by Browser Name"; an empty description becomes the application name as the panel subtitle, which tells panels querying different applications apart. Click **Apply** to put the panel back on the canvas, then **Save** to submit. You can fill in a **version message** (optional); each save produces a new version. ## Panel queries A panel query has the same parts as the Explorer's analytics view: | Part | Description | | - | - | | **Application** | One application, or **Page application** (`$app`, so the panel follows the page's application) | | **Event type** | Sessions, Views, Resources, Actions, Errors, Long Tasks, Vitals | | **Aggregation / measured field** | Count, unique count, average, sum, min, max, P50 / P75 / P90 / P95 / P99; every aggregation except Count needs a field | | **Group-by field / top** | Split by a field and keep the top `5`, `10`, `20` or `50` groups | | **DQL filter** | The Explorer's search syntax, see [Data query syntax](../explorer/data-query) | Group-by fields are limited by field cardinality, exactly as in the Explorer (see [Analytics · Group-by field](../explorer/analytics#group-by-field)): a field that is an event's own ID cannot be grouped by, and any other high-cardinality field takes Count or Sum only, with a time range of at most 7 days. When the result is estimated from a sample, the panel shows a **Sampled** badge and its explanation. ## Dashboard variables The variable bar sits at the top of the dashboard. A new dashboard starts with three field variables, `env`, `service` and `version`, all defaulting to **All** (All means no condition). | Variable kind | Description | | - | - | | Built-in `$app` | The page's current application. Choosing **Page application** in a panel query is the same as using `$app`; only an application's `application_id` can reference it | | Field variable | Candidates are the values of one RUM field; multi-select, All allowed | | Custom variable | Candidates you type in (one per line, `label=value` or a bare value); single or multi-select | * Reference a variable in a panel's DQL as `${name}` or `$name`, such as `env:${env}`. A variable with several values expands to `(a OR b)`; with **All** selected, or with no value, the whole condition holding the reference (`env:${env}`, `-env:${env}`, `NOT env:${env}`, and a value list containing it) is dropped rather than queried as an empty value. * References are replaced in the browser before the query runs, so a panel's DQL has to fill in into a complete statement to return data. * A new panel's filter starts with the dashboard's field variables that its event type has. * Renaming a variable rewrites the queries that reference it; a variable still referenced cannot be deleted until those references are cleaned up. * Variable selections are kept in the link (parameters such as `var-env=...`), so copying the link shares the same selection. A RUM dashboard has just these two kinds of custom variables: field variables that list the values of a RUM field, and custom variables with candidates you type in. Datasource variables and Prometheus / SQL / logs variable queries do not apply and never appear on a dashboard. ## Saving from the Explorer In the Explorer's analytics chart toolbar, **Save to dashboard** turns the current question into a dashboard panel: * The chart type can be **Time series / Table / Stat**; pies and distributions are not supported yet and the button says "Pies and distributions cannot be saved to a dashboard yet". * You can **append to an existing dashboard** (its edit mode opens with this panel appended, and nothing is written before you save), or **create a new dashboard** (a dashboard holding just this panel is created first, then its edit mode opens). * The panel keeps the current time range and filter and picks up the dashboard's `env` / `service` / `version` variables; its title is generated from the query. For the procedure and limits, see [Analytics · Saving to a dashboard](../explorer/analytics#saving-to-a-dashboard). ## Table drill-down Every row of a **Table** panel carries an "Open in Explorer: `field = value`" link. Clicking it opens the Explorer's analytics view with the panel's query and aggregation, its filter narrowed to that row's group value, over the dashboard's current time range. * A table without a group-by has no drill-down link. * When a group value contains characters that a DQL condition cannot express exactly, such as quotes or brackets, the link matches by wildcard instead, and its title is marked "(approximate wildcard match)". ## Version history and restore **Version history** in the dashboard's **⋯** menu opens the version list: * Each save (deleting and restoring included) produces a new version; the most recent **20** versions are kept and older ones are pruned. * The table lists the **version** (`v1`, `v2`, …), its **time**, the **actor** and the **message**, and marks the current version as **current**. While the dialog is open, tick two versions and click **Compare** to diff what each one saved. * Restoring an earlier version saves that version's content as a new version, so the version you restored from stays in the history and you can restore back to it. ## Deleting and the trash * **Delete** in the dashboard's **⋯** menu moves the dashboard to the trash; the confirmation names the dashboard and says it "will be moved to trash and purged after 30 days." * **Trash** on the management page lists deleted dashboards with the note "Deleted dashboards are kept for 30 days and can be restored here."; the **restore** icon at the end of a row puts the dashboard back in the list and in the tab bar. * A dashboard left in the trash for more than 30 days is purged and cannot be restored. RUM dashboards have no folders: they are not grouped, and the management page has no folder tree or cross-folder move. The list can be searched by title. ## Permissions Creating, editing, deleting and restoring a dashboard needs the **Dashboards Manage** permission. Without it: * the **+** menu has no **New dashboard**, and the management page has no **New dashboard** button and no delete or restore entry; * a dashboard has no **Edit**, and its **⋯** menu has no **Clone** or **Delete**; * you can still view every dashboard, change its variables and time range, and inspect its content with **Export JSON** and **Version history**. ## Limits | Item | Limit | | - | - | | Tabs in a dashboard | 10 | | Sections per tab | 10 | | Panels per tab | 30 | | Panels per dashboard | 100 | | Queries per tab | 100 | | Queries per dashboard | 300 | | Variables | 20 | | Candidates per variable | 1000 | | Data links per panel | 5 | | Queries in a time series panel | 1–26 | | Queries in a table / stat panel | exactly 1 | | Markdown size in a text panel | 64 KB | | Title length | 189 characters | | Description length | 1024 characters | | Dashboard definition size | 1 MB | | Default time ranges | `now-15m`, `now-30m`, `now-1h`, `now-3h`, `now-6h`, `now-12h`, `now-24h`, `now-7d` | | Auto refresh options | off, `30s`, `1m`, `5m` | ## Next steps The four analysis dimensions of the built-in Insights dashboard Aggregations, groups, chart types and drill-down The search syntax behind a panel's filter The list and analytics view modes This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.