Skip to main content

FAQ

Why am I am getting 401/ 403 / 500 errors with my ERP connection​

Answer​

Usually this is down to incorrect permissions for the Unit4 User.

The most typical will be:

401- Unauthorized (no credentials given or credentials incorrect)

403- Forbidden (credentials given but no access- default client onuser, menu access, data control)

404- Not found (URL was not available or for SOAP the http / https POST request doesn't match how the service is published e.g. https request to http endpoint)

405- Method not allowed- usually on SOAP and the url hasn't been set up with service.svc on the end in the config.

500- Internal server error (check your web server API / Webservice logfiles)

503- Service unavailable (check your server and firewall config)

My document has "export failed", what do I do?​

Answer​

You should be able to see an error message on the document as to why the export failed.

To re-export the invoice, select the checkbox on the invoice line in the list of invoices (you can select many) and choose the "Back to review Status". We would not advise using "Re-extract" unless necessary as this will incur additional cost.

The AI is not learning​

Answer​

The AI should start to learn your invoices pretty quickly, usually within a few invoices but more complex examples could take up to 10 or so. However if there are particular issues you are seeing over and over then please log this with support (if you use our hosted service) or Rossum (if you contract directly with them) as the team are usually very helpful at resolving the issues.

The AI keeps getting table fields wrong or "double tagging" columns with two fields​

Answer​

We have found that on occasion the AI will identify fields twice, so for example as Net Amount and Gross Amount, and sometimes it will refuse to learn when you remove one of those tags.

The AI engine comes pre-trained to recognize certain fields (full list available at: https://elis.rossum.ai/api/docs/#extracted-field-types). Rossum's recommendation is to remove the "AI engine field ID." This adjustment will allow the engine to bypass its pre-training and instead rely on previously confirmed annotations.

To remove this, find the field in the queue settings in rossum and scroll down to the Captured by AI engine section and remove the selected value from AI engine field ID

Can we stop the OCR extracting extra pages, to save money?​

Answer​

Yes, we can set a effective_page_count to x number of pages, but this will apply to your whole queue. The OCR will then try and extract from the first x pages only.

If the data from the document is not satisfactorily extracted within the first x pages, the invoice will be sent for review.

If the user then selects data from subsequent pages then those pages will become billable.

At this time this queue setting can only be updated via API, so please contact support to request this if necessary

More information about the billing is available here

In the webservice logfile I am seeing System.Xml.XmlException: Data at the root level is invalid. Line 3, position 11.​

Answer​

This is a known issue in some updates of Unit4 MS7 / CR where the import webservice was broken and had issues with the WS_RESTRICTED_HTML parameter settings. Please check the Unit4 community for more information and any hotfixes that might be required.

Where can I see the XML that rossum sent to ERP7 / ERPCR?​

Answer​

You can get the XML from the table acrxmlimport, note you will need to use SQL to convert the blob file to XML to be able to read it.

On MS7 or CR my EI02 or IMS process failed, why does the document still show as exported?​

Answer​

EI02 / IMS failures should be appropriately handled in Unit4 as we can not “see” any errors once we send the data over. The API we’re using only allows us to push the request to be run later, not to run the EI02.