mirror of
https://github.com/BerriAI/litellm.git
synced 2026-09-15 23:31:29 +00:00
21bbba687f
The new global QueryCache.onError already routes every query error through handleError. The query fetch helpers in these hooks also called handleError inline, so each failure dispatched twice; today the second call is swallowed only by the 60s throttle in handleError, which is fragile if that throttle or handleError ever changes. Drop the inline call from the query paths and let the throw reach the global sink, which re-derives the same message. Mutation hooks are left as-is because there is no global MutationCache.onError; removing their inline handler would drop error reporting entirely. |
||
|---|---|---|
| .. | ||
| litellm-dashboard | ||
| Dockerfile | ||
| nginx.conf | ||