Skip to main content
Question

OpenAPICaller fails with Kerberos Authentication on local Data Virtualization service

  • Oct 21, 2026 5:49 AM UTC
  • 1 reply
  • 51 views

featuremichael
Enthusiast
Forum|alt.badge.img+9
Hi everyone,
I am facing an issue with FME Flow 2026.2 where an OpenAPICaller transformer is unable to fetch data from an internal service due to Kerberos authentication failing.
Here are the details of our setup:
  • .The Service: A "Data Virtualization" service hosted internally within our corporate network.
  • The Environment: FME Flow is running locally/on-premise within the same corporate network (Local / Firmennetz).
  • The Problem: The OpenAPICaller cannot authenticate via Kerberos, causing the data retrieval to fail.
Has anyone successfully configured the OpenAPICaller (or underlying HTTP/web connections in FME Flow) to work with Kerberos authentication in a local network environment?
Are there specific environment variables, or FME Flow service account permissions (e.g., SPN / Active Directory delegation) that we need to configure to make Kerberos pass-through work?
Any help or workarounds would be greatly appreciated!
 
Best regards,
Michael

1 reply

nickcumming
Safer
Forum|alt.badge.img+14

Hi featureMichael I had a chat with the team covering Kerberos in OpenAPICaller. It uses Windows integrated authentication, which negotiates Kerberos and only falls back to NTLM if Kerberos isn’t available. Kerberos Authentication is possible with SSO, but how exactly to configure that to not fail in your environment is the challenge here.

A quick way to narrow things down is to run the same OpenAPICaller with Single Sign-On in FME Form on a domain-joined machine. If it works there but not in FME Flow, the Engine service account is the most likely cause.

In FME Flow, the request runs as the Windows account the FME Flow Engine service runs under, not the user who submitted the job. If the service runs as Local System, it authenticates as the server’s computer account, which usually doesn’t have access to the target service. We recommend running the Engine service as a domain service account and granting that account access to your Data Virtualization service. 

Also if your URL is an IP address or a hostname without a matching SPN the authentication will silently fallback to NTLM. Probably not the issue here, but worth noting.


If that doesn't meet you IT requirements, or if you’re still running into issues let me know. It might be worth starting a support case where you can share the exact error or HTTP status from the job log, your FME Flow build, DV service details, and the operating system. From there we can work together to find a solution.