Diagnose API errors

Start by identifying which interface failed: a dashboard action, an HTTPS management request, or a native database connection. Database-driver errors are covered in connection troubleshooting.

Capture useful evidence

Record the UTC time, method/path, HTTP status, sanitized response body, relevant workspace/database UUID and X-Request-ID. Keep request IDs as strings. Redact Authorization, cookies, passwords and connection strings before sharing logs.

Use the public health request to check API reachability separately. Then inspect the failed operation’s reference; different endpoints require different IDs, roles, state and payload fields.

Symptom	What to check	Next action
401 with a developer token	Correct API hostname/header, expiry, revocation, active account and deployed version	Rotate invalid credentials through Settings; never add cookies as a workaround
Dashboard session rejected	Account sign-in/session state	Sign in again through the supported flow; do not export cookies
403	Required scopes, token-supported route, active subscription, role and quota	Use the required scopes or the dashboard for session-only operations
404	Resource UUID, path and trailing slash; membership can hide resources	Verify ID and account/workspace; do not retry guessed identifiers
400 on editor calls	Missing or mismatched workspace_id; missing query; numeric pagination	Send the documented query parameters and body type
400 validation map	Required field, minimum length, enum, CIDR or schedule day	Correct named fields; preserve the full validation body in diagnostics
LIMIT_EXCEEDED	Requested quota or retention	Inspect owner’s plan usage and returned limit
ENTITLEMENT_UNAVAILABLE	Plan verification unavailable	Retry a read later or contact support; do not purchase again blindly
503 with verification_unavailable	Account/subscription verification unavailable	Retry reads with bounded backoff; access stays denied while verification fails
429	Request volume and any response guidance	Apply bounded backoff; avoid parallel retry storms
500/502, timeout or non-JSON gateway body	Service failure or engine collection failure	Keep status/request ID and reconcile state before repeating mutations
HTTP 200 but unavailable/unhealthy/empty data	Body flags and database state	Diagnose the underlying feature; HTTP success is not feature health

First-call mistakes

• Use https://api.lordbase.com/e/v1 as the API root and retain trailing slashes.
• Send JSON with Content-Type: application/json for JSON bodies.
• Use a database resource UUID in management URLs, not the native database name.
• Parse workspace lists as owned_workspaces plus member_workspaces.
• Do not call response.json() on a 204 response.
• Do not assume a successful token-creation response proves token authentication works.

For a create/clone/restore timeout, inspect existing resources and retain any returned ID before retrying. There is no documented idempotency key or rollback guarantee for those operations.