scriptoriumadmin/litellm
1
0
Fork 0
mirror of https://github.com/BerriAI/litellm.git synced 2026-10-10 03:28:53 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 85

fix(router): copy kwargs instead of popping target order

Browse source 
Lint required a specific exception on the empty-order-2 regression.
Provider calls now omit _target_order by constructing a new kwargs dict.
This commit is contained in:
Emerson Gomes 2026-08-31 10:35:10 -05:00
parent bd0b9c78bd
commit f7b1cc1f41
No known key found for this signature in database
GPG key ID: D3DF28AB5D1B5E17
3 changed files with 5 additions and 8 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

6
litellm/router.py
View file

@ -2172,9 +2172,8 @@ class Router:
"messages": messages,
"caching": self.cache_responses,
"client": model_client,
**kwargs,
**{k: v for k, v in kwargs.items() if k != "_target_order"},
}
input_kwargs.pop("_target_order", None)
response: Final = litellm.completion(**input_kwargs)
verbose_router_logger.info("litellm.completion(model=%s)\x1b[32m 200 OK\x1b[0m", model_name)
@ -3194,11 +3193,10 @@ class Router:
"messages": messages,
"caching": self.cache_responses,
"client": model_client,
**kwargs,
**{k: v for k, v in kwargs.items() if k != "_target_order"},
}
input_kwargs.pop("silent_model", None)
input_kwargs.pop("include_fallback_errors", None)
input_kwargs.pop("_target_order", None)
_response: Final = litellm.acompletion(**input_kwargs)

3
litellm/router_utils/fallback_event_handlers.py
View file

@ -412,11 +412,10 @@ async def run_async_fallback(
# LOGGING
kwargs = litellm_router.log_retry(kwargs=kwargs, e=original_exception)
verbose_router_logger.info("Falling back to model_group = %s", mask_sensitive_structure(mg))
kwargs = {k: v for k, v in kwargs.items() if k != "_target_order"} # rebind-ok: next hop must not inherit the previous order target
if isinstance(mg, str):
kwargs["model"] = mg
kwargs.pop("_target_order", None)
elif isinstance(mg, dict):
kwargs.pop("_target_order", None)
kwargs.update(mg)
fallback_depth = fallback_depth + 1
_hop_metadata = dict(kwargs.get(metadata_variable_name) or {})

4
tests/test_litellm/test_router_order_fallback.py
View file

@ -14,6 +14,7 @@ import litellm
from litellm import Router
from litellm.integrations.custom_logger import CustomLogger
from litellm.router_utils.prompt_caching_cache import PromptCachingCache
from litellm.types.router import RouterRateLimitError
from litellm.utils import _get_deployment_order, _get_order_filtered_deployments
# ---------------------------------------------------------------------------
@ -454,13 +455,12 @@ async def test_router_order_fallback_does_not_reselect_order_1_when_order_2_is_f
)
litellm.callbacks.append(drop_order_2)
try:
with pytest.raises(Exception) as exc_info:
with pytest.raises(RouterRateLimitError, match="No deployments available") as exc_info:
await router.acompletion(
model="test-model",
messages=[{"role": "user", "content": "hi"}],
)
assert "success from order 2" not in str(exc_info.value)
assert getattr(exc_info.value, "_hidden_params", {}).get("model_id") != "1"
finally:
litellm.callbacks.remove(drop_order_2)