phoenix/litellm
1
0
Fork 0
forked from phoenix/litellm-mirror
Code Pull requests Activity

fix - cooldown based on exception header

Browse source
This commit is contained in:
Ishaan Jaff 2024-05-17 18:52:45 -07:00
parent 6368d5a725
commit f7a1675337
1 changed files with 12 additions and 5 deletions
Download patch file Download diff file Expand all files Collapse all files

17
litellm/utils.py
View file

@ -8008,11 +8008,8 @@ def _should_retry(status_code: int):
return False
def _calculate_retry_after(
remaining_retries: int,
max_retries: int,
def _get_retry_after_from_exception_header(
response_headers: Optional[httpx.Headers] = None,
min_timeout: int = 0,
):
"""
Reimplementation of openai's calculate retry after, since that one can't be imported.
@ -8038,10 +8035,20 @@ def _calculate_retry_after(
retry_after = int(retry_date - time.time())
else:
retry_after = -1
return retry_after
except Exception:
except Exception as e:
retry_after = -1
def _calculate_retry_after(
remaining_retries: int,
max_retries: int,
response_headers: Optional[httpx.Headers] = None,
min_timeout: int = 0,
):
retry_after = _get_retry_after_from_exception_header(response_headers)
# If the API asks us to wait a certain amount of time (and it's a reasonable amount), just do what it says.
if 0 < retry_after <= 60:
return retry_after