2015-09-21 17:05 GMT+02:00 Mark Thomas <ma...@apache.org>:

> If Jean-Frédéric could share a test that reproduced the problem that
> would be extremely helpful.
>
> It involves running ab on overpowered systems as part of the testing of
OpenSSL, and so now waiting to get the lab back for further testing (the
direct BB "leak" with NIO is confirmed fixed, that was a more serious
issue). It's rather likely there's something slightly wrong since in this
kind of tests there's not supposed to be some unexpected disconnects.

Rémy

Reply via email to