Tenants

/admin/tenants — create, inspect, and manage every tenant on the platform.

Table of contents

  1. The tenant list
    1. Per-tenant actions
  2. Creating a tenant
  3. New-tenant provisioning with LoginLink
  4. Connecting LoginLink per tenant
  5. Console key
  6. Support view
  7. Managed databases
  8. Deleting a tenant

The tenant list

Each tenant is a card showing its name and slug, an Active/Inactive status badge, and a row of counts: databases, roles, app users, keys (every API key/Basic login/issuance key the tenant has), app tokens currently in use (backend-issued tokens for end-users, not team sign-ins), and calls kept (usage-log detail retained for 30 days). Two more badges appear conditionally: a LoginLink connection badge, and a managed- database mode badge (see below). Clicking the tenant’s name opens its active console/app sessions, with a Revoke option per session.

Tenants waiting on a LoginLink self-service request you haven’t approved yet sort to the top of the list, with a banner counting how many are waiting, so none get missed below the fold.

Per-tenant actions

Button Does
Edit Change the tenant’s display name, or flip Active/Inactive.
🔗 LoginLink Open the LoginLink connection modal (below).
👁 Support view Open a read-only, 30-minute view of the tenant’s own console, logged with a reason.
🔑 Console key Issue a console-only sign-in credential for this tenant’s administrator.
🗄 Databases Configure managed-database limits and provisioning mode.
🔑 Access Jump to Access model, pre-filtered to this tenant’s guardrail exception.
Delete Permanently remove the tenant record (only once it holds nothing — see below).

Creating a tenant

+ Create Tenant asks for an id (unique, used internally — never shown in URLs), a name, and a slug (auto-suggested from the name, editable; this becomes part of its sign-in address). If new-tenant provisioning is on, the form also asks for the first admin’s email (required — becomes their LoginLink sign-in) and name (optional).

If LoginLink provisioning is on and succeeds, you’re shown the new organisation’s admin password once — the Close button on that dialog stays disabled until you’ve copied it, since it is never shown again. If provisioning fails, the tenant is still created; a Retry control appears on its card.

A platform-wide switch, shown as a status card at the top of this page: when on, every + Create Tenant also creates that tenant’s own sign-in organisation on LoginLink automatically. The card shows how many tenants exist in total, how many were created via LoginLink, and how many were connected to LoginLink by hand instead.

Turning it on walks a three-step wizard:

  1. LoginLink server — read-only here; set once in Host Settings.
  2. Credential — paste a Client ID and Secret from a LoginLink credential that has only the CreateTenants permission ticked. The secret is stored encrypted and never shown again; closing the wizard without finishing clears whatever you’d typed.
  3. Check and turn on — a live check call to LoginLink confirming the credential can actually create tenants. Nothing is saved until this check passes.

The 🔗 LoginLink button opens a modal whose content depends on whether that tenant is already connected:

  • Already connected — shows the issuer and client id it’s connected as, with a Disconnect option. While you manage the connection, the tenant cannot connect its own.
  • Not connected — first asks who should be in control:
    • The tenant connects it — they link their own LoginLink organisation from their own Access tab; you just turn this on and approve who gets in. A toggle further lets you allow this per tenant.
    • You connect it — handled right here, with two paths:
      • ⚡ Create a new one — CrudAdmin creates a brand-new LoginLink organisation using the platform-wide provisioning credential (needs an admin email + optional name for that org’s first user). Disabled with an explanation if platform-wide provisioning isn’t on.
      • 🔗 Connect an existing one — a four-step guide for finding the tenant slug (from the LoginLink console’s own URL), creating an API-access credential there with Manage apps ticked, and pasting its Client ID/Secret here. An Advanced section lets you override with a full issuer URL for a LoginLink instance on its own custom domain.

Console key

Issues a credential that signs one person into this tenant’s own console (/{slug}/login) and nothing else — no database, no role, so it can never read or write data. Give it a label (e.g. “Alice, tenant admin”) and an expiry (7/30/90/365 days, or never — not recommended). Shown once; copy it before closing.

Support view

Opens the tenant’s own console exactly as they’d see it, read-only, for 30 minutes. You must give a reason (up to 300 characters) before opening it — the tenant can see this reason in their own activity log, so support access is never silent.

Managed databases

Decides whether CrudAdmin may provision real databases on this tenant’s behalf (vs. the tenant only registering databases it already has elsewhere), and how many it may have in total:

  • How many databases can it register? — Unlimited (default) or a specific number, counting every database the tenant has registered, managed or not. Lowering the limit below what it already has doesn’t remove anything — it only blocks registering more.
  • Should CrudAdmin create databases for it?
    • Off — the tenant only registers databases it already has.
    • Ask first — the tenant requests one, and you approve or reject each request from this same modal.
    • Self-service — the tenant creates its own managed databases directly, up to a limit you set.
    • For either of the last two, you also set a name prefix — what the database is actually called on the underlying server (never used in the tenant’s own API calls, which always address it by its own db_key).

Pending requests (in “Ask first” mode) list each requested name and reason with Approve/Reject buttons. Approving genuinely calls out to the real provisioning account (Plesk, cPanel, or a MySQL server you control) — it can take a few seconds, and the button shows a spinner rather than appearing frozen. A request that was approved but failed to actually provision (a name collision, an unreachable account) stays visible with the error and a Retry button, rather than silently vanishing into the audit log.

Deleting a tenant

Deletion first runs a live check of everything the tenant still holds — app users, keys of every type, databases, roles, active app tokens, current team sign-ins. If it holds anything at all, nothing is deleted — you get a count of exactly what’s in the way, and need to clear it first. Only once a tenant holds nothing does the delete button activate, and even then you must type the tenant’s exact id to confirm. This cannot be undone. If the tenant was connected to LoginLink, deleting it here does not remove its LoginLink organisation — that stays and needs removing there separately, if desired.


Copyright © CrudApi. Distributed under an open license.

This site uses Just the Docs, a documentation theme for Jekyll.