Hi Francesco, we received the support save.
Some issues detacted in regards:
1. Timeouts from several backends. You could increase the backend timeout for those cases
HTTP/1.1: Connection timed out (45.009 secs)
2. All backends from certain farms come down everyday at 7:00am, probably due to a scheduled task.
3. Broken connections to backends due to client timeouts
error read from XXX: Connection timed out
backend YYYY, (…..) error copy server cont: Broken pipe
When a client timeout is detected, then the proxy closes the backend side. You can try to increase client timeout and see if this palliates the issue.
Cheers.