Skip to main content
Question

Missing o_celonis_VendorAccountCreditItem relation in Duplicate Invoice Checker (Catalog 3 / OCPM)

  • September 8, 2026
  • 2 replies
  • 33 views

m.alzate
Level 2
Forum|alt.badge.img+1

Hi Everyone,

I am currently setting up the Duplicate Invoice Checker (Catalog 3) app using the Accounts Payable Catalog (v3.0.1).

After enabling the Accounts Payable catalog in Objects & Events, the objects VendorAccountCreditItem and VendorAccountDebitItem were not created (only VendorAccountItem exists). When running the autogenerated job auto_generated_dic_combine_vendor_credit_debit_account_items, I encounter the following error:

Relation "1a2b3c4d5e_OCDM.o_celonis_VendorAccountCreditItem" does not exist. Please review your query and try again.

Environment Details:

  • Accounts Payable Catalog: Version 3.0.1

  • App: Duplicate Invoice Checker (Catalog 3 / OCPM)

  • AI Annotations / Signals: Enabled and in "Ready" status

Has anyone encountered this issue when migrating or deploying the OCPM version? Is there an updated perspective or script that replaces the VendorAccountCreditItem reference with VendorAccountItem? Any guidance on how to resolve this missing relation would be greatly appreciated!

Thanks in advance!

2 replies

gagan1
Level 12
Forum|alt.badge.img+6
  • Level 12
  • September 8, 2026

Hi ​@m.alzate ,
 

First thing worth checking: the exact table name behind the VendorAccountItem you can see. If it is o_DuplicateInvoiceChecker_VendorAccountItem then it is the app's output, not a catalog replacement. That job is what creates it, by combining VendorAccountCreditItem and VendorAccountDebitItem from the standard Accounts Payable perspective. So the two missing objects are its inputs, and they have to exist on the AP side first.
 

Also worth knowing: "relation does not exist" is different from empty. An object type with no data still has a table, and the job would return 0 rows instead of failing. A missing relation means it was never materialised in that environment.

The usual cause of that is development vs production. Dev tables carry an extra
t_ prefix, so if the AP catalog is published to development only, yours are t_o_celonis_VendorAccountCreditItem and the production job looking for o_celonis_VendorAccountCreditItem finds nothing. Same split on the jobs,

test:ocpm-data-job vs ocpm-data-job.

So I would check, in this order :)
 

  1. Objects and Events > Catalog > Accounts Payable. Check Enable process is on, and that a data connection is actually added under it with the Add button. Enabling alone does not add the transformations.
  2. Create a version, deploy to production, then run ocpm-data-job.
  3. Check whether o_celonis_VendorAccountCreditItem and o_celonis_VendorAccountDebitItem now exist without the t_ prefix.
  4. Then run auto_generated_dic_combine_vendor_credit_debit_account_items, scheduled after ocpm-data-job.

I would not repoint that job at VendorAccountItem. That is its own output table, and the AI Annotation reads it downstream.

If the table you see turns out to be o_celonis_VendorAccountItem instead of the DIC one, say so, that would mean something different and worth a closer look.

Docs --» [ they may help you ]
https://docs.celonis.com/en/object-centric-process-mining-database-tables.html
https://docs.celonis.com/en/installing-the-object-centric-duplicate-invoice-checker.html


m.alzate
Level 2
Forum|alt.badge.img+1
  • Author
  • Level 2
  • September 8, 2026

Thanks, that's a great breakdown. To confirm: the table I see is indeed o_DuplicateInvoiceChecker_VendorAccountItem, so it's the app's own output table, not a catalog replacement. It exists but is empty, which lines up with its inputs (VendorAccountCreditItem/DebitItem) never being materialized. These two objects don´t exist on my Data Pool.

On the dev vs. production / t_ prefix theory: I went through steps 1 and 2 already. The AP catalog has "Enable process" on, and a data connection is added under it (not just enabled). I created a new version, deployed to production, and ran ocpm-data-job. Nothing changed after that. Same error appeared.

One more data point: I also tried disabling "Skip missing data" on the connection, re-publishing, and re-running the job, just in case a source mapping was being silently skipped. Same result, no change, and no explicit error in the job log either.


What else can I do?