fix(envoy_ai_gateway): publish inference.<domain> in keycloak mode only - #12
Open
Mushtaq-BGA wants to merge 1 commit into
Open
Mushtaq-BGA wants to merge 1 commit into
Mushtaq-BGA wants to merge 1 commit into
Conversation
The inference-external HTTPRoute forwards inference.<domain> straight to the ai-gateway, bypassing LiteLLM. It was created whenever ai_gateway_create was set, so litellm mode exposed a direct, unauthenticated path around LiteLLM's virtual-key auth, budgets and logging. Gate creation to keycloak mode, where the gateway JWT policy guards it, and remove any route left behind by a prior keycloak install on a switch to litellm/none. LiteLLM still reaches ai-gateway over its own internal ClusterIP service, so the model path is unaffected.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Problem
The
inference-externalHTTPRoute forwardsinference.<domain>straight to the ai-gateway, bypassing LiteLLM. It was created wheneverai_gateway_createwas set, regardless of auth mode — so litellm mode exposed a direct, unauthenticated path around LiteLLM (virtual-key auth, budgets, rate limits, Langfuse logging).Change
inference-externalcreation to keycloak mode, where the gateway JWT policy guards it.LiteLLM still reaches the ai-gateway over its own internal ClusterIP service (per the documented litellm request flow), so the model path is unaffected. In litellm mode
litellm.<domain>is the single external inference endpoint.Companion
Pairs with intel/enterprise-ai-solutions#17 (gateway auth scoped to keycloak; litellm gateway is TLS+routing only).
Test
yamllintclean;auth_provideris the same global var already used by the langfuse/litellm/llm_services roles.