Common questions encountered while using Connector, and how to resolve them.
Connections and permissions
A RAM user cannot use Connector
A RAM user must first be authorized by the Alibaba Cloud account. Ask the Alibaba Cloud account to grant the user the required data connection permissions in the RAM console, then try again.
An error says that permissions for this module are missing
The current account lacks the operation permissions for the corresponding module. If you are a RAM user, contact the Alibaba Cloud account to complete the authorization. If the Alibaba Cloud account sees this message, confirm whether the related service has been activated.
An OSS bucket does not appear in the drop-down list
Check the following in order:
- Whether the service-linked role has been initialized. Click Authorize now in the connection dialog box to complete it.
- Whether the bucket tag is correct. The bucket needs
bailian-datahub-accesswith the valueread. - Whether the current account has read permissions on the bucket.
Is there a security risk in uploading confidential files to a connector
Imported files are available only to users in the current workspace. The platform does not use them for commercial purposes or make them public. For confidential data, we recommend that you:
- Use a dedicated workspace for isolation instead of sharing one with other teams.
- Strictly control who is a workspace member and how API keys are distributed.
Auth configs
The connect dialog only has one dropdown
This App requires an auth config first. Expand the Select auth config dropdown. If it shows that there is no config yet, click Create auth config below it to create one. For the flow, see Create an auth config.
A config I already created does not appear in the dropdown
An auth config is bound to one App. A config created for Yuque does not appear in the Salesforce dropdown. Create a separate one for the current App.
Can one config be used by several connectors
Yes. Multiple connectors under the same App can share one auth config. You only need separate configs when the two accounts have different credentials.
Does deleting a connector delete its auth config
No. The config remains in the Auth configs list and can be reused by a new connector.
What do I do when the provider rotates its secret
Create a new auth config with the new credentials, rebuild the connector with that config, and delete the old connector once you have confirmed the new one works. Do not delete an OAuth app that is still in use on the provider side, because that breaks every related connector immediately.
MCP integration
A 401 is returned
The authentication header is missing, or the API key is invalid or expired. Confirm that the header format is Bearer <API-Key>, and check the key status in the console. After you replace a key, you must restart the client.
A 404 is returned
The path in the MCP endpoint is incorrect. Confirm that the path is /api/v2/connector/mcp, and note that the version number is v2.
The client shows as connected but the tool list is empty
There is no connected app in the current workspace yet. Create a connector on the Apps page first. Also confirm that the workspace ID in your configuration matches the workspace where the connector was created.
Tool calls suddenly fail
The credentials for the corresponding connector may have expired. Check the connector status on the App details page. If it shows as Expired, you need to authorize it again.
File parsing
Parsing takes a long time or occasionally times out
During peak request hours, parsing can take several hours. High concurrency adds further delay because of resource queuing.
What to do: wait patiently or retry later, and do not repeatedly submit the same batch of files. Schedule large imports outside peak hours whenever possible.
Imported files are no longer visible
Only files imported within the last 90 days can be viewed. Beyond that time range, files are no longer displayed, but they are not deleted.
JSON, CSV, and YAML files cannot be imported
Direct import of these three formats is not supported at this time. Convert them to XLSX or XLS with an appropriate tool first, then import them through the table connector.
Scanned documents are parsed poorly
Select custom parsing settings during import and switch to a parsing method intended for scanned and image-based documents.
OSS import
Importing OSS files returns error 10041495
This is usually an authorization or path issue. Confirm that the service-linked role has been initialized, that the bucket carries the bailian-datahub-access=read tag, and that the object path includes the bucket name.
Why are OSS outbound traffic fees incurred
Reading data from OSS generates outbound traffic fees on the OSS side, charged by OSS according to its own billing rules and calculated separately from your Connector usage. This cost becomes noticeable when the same bucket is read frequently.
To control costs, consider importing the data into platform-hosted storage once instead of reading it from OSS in real time on every request.
How should the object path be written
The path must include the bucket name:
| Format | Meaning |
|---|---|
my-bucket/docs/ | Import all files in that directory |
my-bucket/docs/foo.md | Import a single file |
/docs/ | Incorrect, the bucket name is missing |