Server settings

The Server group in Settings holds the server's read-only deploy status and the defaults every project starts from. Only an administrator can open these pages.

Overview

Open Settings → Server → Overview to see the server's deploy status. Every value here is read-only. An administrator sets secrets — sign-in credentials, mail credentials, connector tokens — only through environment variables. The editable defaults live on the General page.

Value	What it tells you
OIDC SSO	Whether an SSO provider is on the sign-in page.
LDAP / AD	Whether directory sign-in is on, and whether a bind account is set. Click the row to open Directory.
Email	Whether a mail sender is set up. Click the row to open Email.
TOTP MFA	Whether the server allows TOTP multi-factor sign-in.
AI provider	The provider assigned to the chat model role, or "Off". Click the row to open AI.
Attachment storage	The type of the default storage host, for example s3 or filesystem. Click the row to open Storage.
Background workers	Whether background workers run inside this server process. Off can mean they run in a separate process — check Monitoring.
Connectors	How many optional connector plugins have their environment variables set, out of how many exist. Click the row to open Plugins.

General

Open Settings → Server → General for the defaults every project starts from. A project can override each one under its own project settings.

Setting	What it does	Default
Workflow transition enforcement	Off: any user can move a work item to any state. The transition list is ignored. Guarded: a move with a defined transition must meet its conditions and approvals. A move with no defined transition stays allowed. Strict: a move needs a defined transition. The server blocks an undefined move, and a defined move still checks its conditions.	Off
Waiting-for-release state	The name of the state a merged pull request moves a work item to. This state belongs to the done category, so a report counts the day the work finished, not the day someone published a release.	Empty (the release pipeline is off)
Shipped state	The name of the state a release sweep moves a work item to when someone publishes a version. The server records the release on each work item it moves.	Empty (the sweep does nothing)
Story points	Turn on estimating a work item in story points, 0–999 with one decimal, next to time tracking.	Off
CSAT surveys	Email the requester a one-click satisfaction survey when their work item resolves.	Off

Set both the waiting-for-release state and the shipped state to turn on the release pipeline for a project. Leave either one empty to turn that part off.

Page spaces

Open Settings → Server → Page spaces to manage the server's page spaces. A space holds one page tree. Managing a space needs the page.manage permission atom in it.

To create a space:

1. Type a name under New space name.
2. Type an optional description.
3. Click Create space.

A space is a permission scope. Who can read, write, and comment in it depends on a role granted there. Nobody holds that role automatically.

Grant it in one of two ways. Grant it instance-wide under Settings → Roles: the built-in Viewer role carries page.read, and the built-in Member role carries page.write. Or grant it for one space only, from this page.

To grant access to one space, click the shield icon next to it. This opens the space's access panel:

• The panel lists every role grant scoped to this space only. An instance-wide role also applies here but does not appear in this list, because revoking it from this list would remove it everywhere.
• Click Grant role to add one. Select a person or a team, then select a role.
• Click the X next to a grant to revoke it.

To make a space public:

1. Click the pencil icon to edit the space.
2. Turn on Public.

Anyone with the link can then read the space's pages with no sign-in. Copy the link from the field that appears once Public is on. Archived pages stay hidden even in a public space. Use an external URL for an image in a public page — an attachment link on the page still needs sign-in.

To remove a space, click the trash icon. If the space still holds pages, the server asks you to confirm removing the space and every page in it together. This cannot be undone.

Backups

Open Settings → Server → Backups for scheduled and on-demand backups of the database and attachments.

The server needs the pg_dump and pg_restore tools at startup. Without them, it refuses to start. An administrator can set RADD_BACKUP_TOOLS_OPTIONAL to let it start anyway, with backups off.

The server encrypts every backup at rest with a key file kept outside the backup directory. Back up that key file separately. Without it, nobody can restore a backup.

The status card shows the backup directory, its free space, the next scheduled run, the encryption key, and the installed pg_dump/pg_restore versions. A problem with any of these appears below the card.

Schedules

Click Add schedule, then set:

1. A name.
2. When it runs — interval, daily, weekly, monthly, or cron.
3. How many backups to keep, by count, by age in days, or both. Leave both empty to keep every backup.
4. Whether to include attachments as well as the database.

Click Disable to turn a schedule off without removing it, and Edit to change it. Click the trash icon to remove a schedule. Existing backups it already made stay on disk.

Available backups

The table lists every backup on disk: when it was taken, its kind (scheduled, manual, or uploaded), its size, its contents, and who made it. A shield icon marks an encrypted backup.

• Click the download icon to download the file.
• Click Upload to add a backup file taken elsewhere.
• Click Back up now to take one immediately. This needs a writable backup directory and working tools.
• Click the restore icon to restore a backup (see below).
• Click the trash icon to remove a backup file from disk. This cannot be undone.

Restoring a backup replaces every row in the instance. The server takes a safety backup first, and Radd is unavailable while the restore runs.

To restore a backup:

1. Click its restore icon.
2. Type the database name to confirm.
3. Click Restore.

If a restore fails partway, Radd stays unavailable until an administrator restarts the server.

Monitoring

Open Settings → Server → Monitoring for the server's live health. The page refreshes every few seconds while it is open.

• Database shows whether the server can reach the database, its PostgreSQL version, its size, and its active connection count.
• Contents (approximate) shows a row count for each major table — projects, issues, users, comments, worklogs, pages, attachments, events, and search index rows. Postgres's own live-tuple estimate supplies each count, so it is instant at any size and approximate, not exact.
• Semantic index appears only once the embeddings model role has a provider assigned. It shows how many issues and how many pages the server has embedded, out of the total, with a note while a backfill is still in progress.
• Background workers lists every background consumer, what it does, its backlog, when its position last moved, and its status.

Status	Meaning
OK	No backlog, or the backlog is not growing.
Catching up	There is a backlog, and the worker's position moved within the last two minutes.
Stalled	There is a backlog, and the worker's position has not moved for more than two minutes.

A stalled worker means work is waiting, and nothing is processing it. Check that the background-worker process is running. On a server that splits web and worker processes, that is a separate process from the one serving this page. Check its logs for the error. The server stores each worker's position, so once the worker runs again, it continues from where it stopped.

Plugins

Open Settings → Server → Plugins to see every plugin the server knows about, and to turn one on or off.

Each row shows the plugin's name, its version, a state badge, and a short description. A connector plugin also shows a Configured or Not configured chip, based on whether its environment variables are set.

• A core plugin carries a lock icon marked Core. The server never lets you turn it off.
• Click Disable to turn an optional plugin off, or Enable to turn it back on. The change takes effect at once: the server mounts or unmounts the plugin's endpoints, events, permissions, and navigation with no restart.
• If a plugin that is on depends on this one, the server refuses to turn it off. Turn off the dependent plugin first.
• A plugin that owns a settings page links to it from its own row, for example AI, Directory, or Import from Jira. This link disappears once you turn the plugin off.
• A plugin that contributes its own toggles shows a chevron instead of an icon. Click it to open that plugin's own instance-wide settings inline. Turning one of these off applies to everyone, and also removes it from each person's own per-profile toggle.

Audit log

Open Settings → Server → Audit log for the append-only trail of every attributable change on the server, newest first. Reading it needs the global.manage permission atom.

Each row shows five columns:

• When the change happened.

• Who made it, or "System" for an automated action.

• Action — the event type, for example page.deleted.

• Entity — the kind of record it changed.

• Details — a short summary of what changed.

• Type in the search box to search the action's event type or its payload text.

• Select an entity type from the dropdown to narrow the list: items, comments, worklogs, related links, version control, states, fields, roles, users and members, cycles, releases, saved views, automations, or intake forms.

• The list pages 50 rows at a time.

Use the audit log to answer "who did this, and when". It is the record of every write the server made, not only the ones a person sees on screen.

---

Mirrored from project.radd-hq.com on 2026-09-12. Documentation is written there; this copy is regenerated by scripts/publish_wiki.py and hand edits do not survive it.