Fledge v0.3.1.1

A redesign of the whole panel. Same features, same API — a clearer, faster interface that shows what your servers are actually doing.

Highlights

• Live resources on every server page. CPU and memory are drawn as live graphs from the node's samples, next to disk use. The values update in place as new samples arrive.
• Live usage in server lists. The Servers page and the Overview show each running server's current CPU and memory against its limit, not just the configured allocation.
• A console that keeps its history. Log output now survives switching between tabs, highlights warnings and errors, and offers "Jump to latest" when you scroll back. Up/Down recalls previous commands.
• One place for each problem. The Overview lists offline nodes and failed servers once, with the error from the failed job and a direct link to it.
• Search everything with Ctrl K. The command palette finds pages and servers, with keyboard navigation.

Changed

• New visual design: a warm dark theme with a single accent taken from the Fledge mark. Colour is used only for state (healthy, warning, failed, working).
• Motion throughout: sliding selection in the sidebar, tabs, and filters; drawers, dialogs, and menus ease in; numbers roll when they change. All motion respects the system "reduce motion" setting.
• Action results (restarts, saves, backups) appear as short notifications instead of messages that stay on the page.
• Sizes and limits are shown in GB and cores everywhere, including the forms for new servers, nodes, and server resources.
• Nodes show reserved memory split by the servers on them; hover a segment to see which server it is.
• Nodes that have never connected now read "Waiting for agent" with a Connect button, instead of "Offline". Adding a node opens the connector for it straight away.
• The agent connector command is shown one option per line, ready to copy.
• Activity is grouped by day with exact times, shows names instead of IDs, and hides sign-ins unless you choose "Everything".
• Schedules, collaborators, recovery codes, and API tokens are listed first; new ones are created in a dialog. Schedule intervals accept minutes, hours, or days.
• Server jobs have their own tab; repeated identical jobs are collapsed into one row.
• Start and Stop share one button that follows the server's state. Confirmation dialogs name the action and the server.
• The sign-in page is a single focused form, and two-factor codes are entered in the same step.
• Tables, the console, and server pages are reworked for phones.

Fixed

• Creating a schedule always failed with a server error (inconsistent types deduced for parameter $4). Schedules can now be created.
• Backup sizes are formatted correctly when the database returns them as large integers.
• Live log lines are no longer duplicated after the console reconnects.
• A single failed background refresh no longer replaces an open server page with an error.
• File transfers stop polling once you leave the Files tab.

Removed

• Unused interface components and the recharts dependency.

Upgrading

No database or configuration changes are required. Update as usual from Updates in the panel, or with update.sh / update.ps1 on the host. APP_VERSION in .env.example and the Compose default are set to 0.3.1.1.