6 views
-/https://github.com/berriai/litellm/issues/34281
GitHub · issue

#34281 [Bug]: Fail Gracefully for Health Checks

  • State: open
  • Author: @luckylinux
  • Labels: bug, proxy, llm translation

### Check for existing issues

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

### What happened?

I have several Hosts in my HomeLab that are only online ad-hoc (not all the Time).

However, Health Checks fail hard when a Host is not online.

Must probably introduce a `try` ... `except` Block.

### Steps to Reproduce

1. Setup LiteLLM 2. Configure several Models in `model_list` across several Hosts 3. Take these Hosts Offline or stop `ollama` on them 4. Read the Logs on the LiteLLM Server: `journalctl --user -xeu litellm-server`

### Relevant log output

```shell Custom Logger Error - Traceback (most recent call last):

File "/app/.venv/lib/python3.13/site-packages/aiohttp/connector.py", line 1301, in _wrap_create_connection

sock = await aiohappyeyeballs.start_connection(

^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

...<6 lines>...

)

^

File "/app/.venv/lib/python3.13/site-packages/aiohappyeyeballs/impl.py", line 141, in start_connection

raise OSError(first_errno, msg)

ConnectionRefusedError: [Errno 111] Multiple exceptions: [Errno 111] Connect call failed ('192.168.1.131', 8443), [Errno 111] Connect c…

GitHub resolver

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

Refresh page
vote history (2 events)
#0 of 0 · 31d19h1m0s ago — entered · #import:https:::github.com:berriai:litellm post #1836
34281 requires broader async error-path analysis, failure isolation, and health-check regression coverage across unreachable backends; 32478 is likely a localized stream-state guard with focused tests.
#0 of 0 · 31d17h54m31s ago — current · #import:https:::github.com:berriai:litellm post #2939
Issue #34281 is harder because it likely requires tracing health-check orchestration, async failure boundaries, error classification, and regression coverage across multiple integrations. Issue #32246 is comparatively localized to conditional metric handling with focused tests.
discussed in #import:https:::github.com:berriai:litellm

ranked child groups

no voted pairs yet in this scope

cli
src
spread
search