### Check for existing issues

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

### What happened?

`general_settings.allow_requests_on_db_unavailable: true` is the HA switch that keeps the proxy serving when the database is unreachable, and it works as advertised at the request layer. `PrismaDBExceptionHandler.handle_db_exception`

https://github.com/BerriAI/litellm/blob/61c7e706dd24db17131053445c36bd00ad8fa228/litellm/proxy/db/exception_handler.py#L262-L262

and

https://github.com/BerriAI/litellm/blob/48b5a5a0cc5a694a11219416ee0b6eb6e620e74e/litellm/proxy/auth/auth_exception_handler.py#L67-L67

both consult it and fail open

`/health/readiness` never consults it. When a Prisma DB is configured but unreachable, both `_resolve_public_readiness_db`

https://github.com/BerriAI/litellm/blob/5f864c83ce7b7586e2524f610dd7dc1712f184ac/litellm/proxy/health_endpoints/_health_endpoints.py#L1534-L1548

and

https://github.com/BerriAI/litellm/blob/5f864c83ce7b7586e2524f610dd7dc1712f184ac/litellm/proxy/health_endpoints/_health_endpoints.py#L1457-L1458

set HTTP 503 unconditionally. There is no `general_settings` key that turns this off

With the ship…