Troubleshooting and FAQ (P0 Scenarios)

Use this guide to diagnose the most common issues across the MOI console, AI Studio, Genesis, and MatrixOne instances. For exact HTTP status codes and JSON error codes, consult the Error Codes and Status Reference.

1. Sign-in, invitations, and workspace permissions

Symptom

Likely cause

What to check

Missing buttons such as Create Workflow, Create Knowledge Base, or user management

Your role in the current workspace lacks that global or object permission

Ask a workspace administrator to check your role assignments under Role permissions

Account invitations blocked after 10 users

Account-level registration invitations (default cap 10 per user) are separate from inviting existing users into a workspace

To collaborate inside a workspace with existing MOI users, invite them in User management; see Account management

Credential creation or model calls prompt for identity verification

Your deployment enforces real-name verification

Verify using your currently verified phone number and real name; see Complete real-name verification

Operations on another workspace return 404 / ErrNotFound

Wrong workspace active in the top bar or mismatched X-Workspace-ID header

Copy the exact target ID from the Workspace switcher and retry

2. API authentication and credential errors

Symptom

Likely cause

What to do

AI Studio request fails with ErrMultipleAuthCredentials

Multiple credentials sent together—commonly browser login cookies stored automatically by Postman, requests.Session, or browser fetches

Send only X-API-Key (and X-Workspace-ID when required), and disable or clear cookies for API-key clients; see Troubleshoot failed calls

Sending Authorization: Bearer to AI Studio returns 401 / expired session

Mixing Genesis and AI Studio header conventions

Use Authorization: Bearer for Genesis, and X-API-Key for AI Studio

Genesis request returns 402 insufficient_credit

Account Credit balance is insufficient

Recharge via Top Up Credit and confirm the balance updates before retrying

Model calls or published dynamic services return 429

API Key / service-account RPM/TPM/concurrency limit or upstream provider throttling

Inspect the Retry-After header and retry with exponential backoff, or check allowed models in List models

3. Data ingestion, connectors, and knowledge bases

  • REST API or storage connector verification fails: Confirm authentication type, credential validity, reachable URL/endpoint, and pagination behavior in Connect a REST API.

  • Load task stalls or fails: Expand task run details under Load tasks to verify source encoding, Catalog table schema mapping, and whether the target compute instance is ready.

  • Knowledge base Q&A missing citations or NL2SQL cannot find tables: Confirm document ingestion/indexing completed cleanly (Create knowledge bases and configure sources), and confirm custom SQL tables are properly registered in Catalog (SQL editor).

4. Compute instances, managed sandboxes, and billing

Symptom

Likely cause

What to do

Compute instance fails to activate or start

Pre-activation Credit check failed due to insufficient balance, or custom Python dependency validation failed

Verify Credit balance in View Costs and Usage; for custom Python packages, verify requirements.txt syntax and allowed index URLs in Compute instances

Managed runtime stays in Waiting or returns SANDBOX_QUOTA_EXCEEDED

Workspace reached its managed sandbox quota or compute pool queue

Delete unused managed environments or ask a platform administrator to raise the workspace sandbox quota

Public network call inside managed sandbox is blocked

Either the tool was not enabled or domain:port was omitted from the network allowlist

Allow both the tool (such as pip, git, or curl) and the target domain:port in Terminals, networking, and maintenance

Continuous compute charges while no chat is running

A managed runtime environment remains ready and bound

Managed environments accrue continuous per-minute charges from Ready until deleted and stopped (unlike temporary in-chat ephemeral sandboxes); see How compute instances and sandboxes are billed

5. IM channels and agent integrations

For Feishu, WeCom, Slack, or Telegram webhook verification failures, signature errors, or missing replies, follow IM integration troubleshooting.

6. MatrixOne database instances and billing ledger boundaries

  • Connection refused or timed out: Verify the instance is running, confirm endpoint host/port and TLS settings, and add your client egress IP to the instance allowlist (Database instances).

  • SQL permission errors: Sign in with explicit username:role_name context and confirm GRANT privileges (Permissions: MatrixOne).

  • Database instance charges not in Credit ledger: MatrixOne CU, storage, and renewal invoices are tracked in the instance console ledger; see Instance Billing.

Before opening a support ticket

If the checks above do not resolve the issue, open Feedback Assistant in the console or follow Support and feedback with these diagnostics (never share raw passwords, AK/SK, or access tokens):

  • Workspace ID, affected resource name, and exact timestamp.

  • Relevant load task ID, workflow run ID, conversation ID, SQL Query ID, Genesis Request ID, or top-up order number.

  • Full HTTP status code, JSON error code, and redacted screenshots.

Last updated on