https://issues.apache.org/bugzilla/show_bug.cgi?id=56119
--- Comment #20 from Sebb <[email protected]> --- Thanks. That shows the issue is something to do with connection re-use. However I don't understand why dropping the staleCheck should affect 50% of requests. Or indeed why HC 3.1 should be 100% successful, but perhaps that does retries - I don't know without looking. I've just had another look at the error on the screenshot, which shows "Target Server failed to respond", yet the elapsed time seems to be close to sero. This suggests that the request may be failing before it gets to the server. Is there a server against which we can test? Or can you run a very short test showing the errors with full wire logging (or using WireShark)? -- You are receiving this mail because: You are the assignee for the bug.
