support session pinning for consistent model selection in routing (#813)

This commit is contained in:
Adil Hafeez 2026-03-13 17:32:32 -07:00
parent 785bf7e021
commit 46a5bfd82d
No known key found for this signature in database
GPG key ID: 9B18EF7691369645
8 changed files with 406 additions and 3 deletions

View file

@ -415,6 +415,14 @@ properties:
type: string
model:
type: string
session_ttl_seconds:
type: integer
minimum: 1
description: TTL in seconds for session-pinned routing cache entries. Default 600 (10 minutes).
session_max_entries:
type: integer
minimum: 1
description: Maximum number of session-pinned routing cache entries. Default 10000.
additionalProperties: false
state_storage:
type: object