Test sign-ins and Connection Doctor
Test a connection, inspect checks and send supported fix instructions to IT.
Test the customer’s real SAML or OIDC configuration before activation. A test runs the normal broker security checks, records the result and displays a result page. It does not issue your application an authorization code.
Start and inspect a test
Section titled “Start and inspect a test”Management API
Section titled “Management API”Authenticate with the environment secret key. Start with POST /, then read GET /.
Hosted portal
Section titled “Hosted portal”Your customer’s IT admin starts a test sign-in from the connection’s page in the setup portal and sees the result there.
Open the returned test_url before its expiry. The sealed ticket is single use and bound to the connection. A new test needs a new URL. The newest recorded test determines the result: pending, succeeded or failed, with checks and any catalogue error codes.
A successful test is processed by the event worker, which marks the connection tested. Activation can remain pending until that dispatch completes. Changing identity-provider configuration can invalidate a prior test; rerun it before activation. The hosted test IdP is still unavailable: use a customer-controlled test identity provider.
Read Doctor instructions
Section titled “Read Doctor instructions”For a supported failure, the test result (in the API response and on the portal’s test screen) includes doctor: the code, detected identity-provider flavor, cause, IT administrator message, steps and fix_guide_id. The check’s guide ID points to the same diagnosis. Unsupported failures can have checks and codes without Doctor instructions.
Doctor supports fixes for recognized configurations; it does not repair the identity provider or bypass a failed check. Treat fix_guide_id as an identifier, not a guaranteed published documentation URL. Provider-specific setup guides require their own verification. Start with the error catalogue and compare the instructions with your configured entity ID, ACS URL or redirect URI.
Send instructions from the portal
Section titled “Send instructions from the portal”From a failed test in the setup portal, the IT admin can email the fix instructions to another IT contact of the same organization. Only the latest failed test with supported instructions can be sent; a passed or unsupported failure cannot.
Sending is limited to once per connection and contact in ten minutes. The mail contains the connection, failure code and fix steps, without the signing-in user’s email or attributes. Mail configuration or delivery failure returns service unavailable; a confirmed send means the mail was accepted, not that the recipient has read or acted on it. The management API has no Doctor-send operation.
Rotation progress warnings
Section titled “Rotation progress warnings”SP signing-key rotation reports its own pending reasons and supported Doctor output. Missing acknowledgement, an unfinished 720-hour wait, a missing or failed next-key test, and missing confirmation keep the current key in place. Provider submission alone never starts the wait.
A rollback key retained for more than 90 days produces a retirement warning. If its age cannot be verified, Doctor reports that uncertainty instead. Review the retained rollback signing key and retire it only when it is no longer needed. A warning does not activate, retire or destroy a key. Destruction and staff warnings for unreferenced keys are not implemented.