6 views
-/https://github.com/berriai/litellm/issues/28003
GitHub · issue

#28003 [Bug]: request to Vantage returns "422 Unprocessable Entity"

  • State: open
  • Author: @bubulmet
  • Labels: bug, proxy

### Check for existing issues

- [x] I have searched the existing issues and checked that my issue is not a duplicate.

### What happened?

I set up Vantage integration as described [here](https://github.com/BerriAI/litellm-docs/blob/main/docs/observability/vantage.md). LiteLLM sends the request, but receives 422 error, which means the payload has wrong format.

I tried to upload [dry-run](https://github.com/BerriAI/litellm-docs/blob/main/docs/observability/vantage.md#2-preview-data-dry-run) output as a CSV file via Vantage UI and got the following error: "ServiceName can't be blank".

So I think the problem is that LiteLLM sends to Vantage all the requests, including the ones that are not LLM requests. The field `model_group` (which is translated to `ServiceName` for Vantage) is blank for such requests, and it causes problems.

I see following possible solutions: - filter out all the records that have blank `model_group` (but there still might be some cost, so it's not the best solution) - put some placeholder into `ServiceName` field for the non-LLM requests, so it is not blank for Vantage

### Steps to Reproduce

1. set up Vantage integration 2. make any non-LLM request to LiteL…

GitHub resolver

Import GitHub neighbors on demand. Results are saved as system ingests.

Refresh page
vote history (2 events)
#0 of 0 · 31d18h44m27s ago — entered · #import:https:::github.com:berriai:litellm post #2123
28003 requires cross-cutting observability-pipeline investigation, external API contract validation, and careful handling of nonstandard events, making regression risk higher; 29066 is comparatively localized to aligning a persistence/model contract with targeted auth-path tests.
#0 of 0 · 31d17h54m25s ago — current · #import:https:::github.com:berriai:litellm post #2944
The left issue requires coordinated protocol translation and round-trip preservation across multiple request/response schemas, provider-specific tool semantics, replay behavior, and regression coverage. The right issue is comparatively localized to observability payload validation and filtering/defaulting, with a narrower implementation and test surface.
discussed in #import:https:::github.com:berriai:litellm

ranked child groups

no voted pairs yet in this scope

cli
src
spread
search