[Bug 192889] accept4 socket hangs in CLOSED (memcached)

[bugzilla-noreply at freebsd.org]

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=192889

--- Comment #4 from Arne Claus <arne.claus at trivago.com> ---
As we got to reproduce a similar problem using our test on FreeBSD 9 (as
reported in the memcached ticket) I currently think it might be something like
a race condition in the application, too.

What's strange though is that the responsible code looks mostly single threaded
with the exception of the socket being used with asynchronously. So
theoretically it could be a race in accept(), too.

The only thing I could think of on memcached's side is that there is a race
with libevent so that close and open commands are somehow in the wrong order.
But I don't know the code well enough to pin that down. Furthermore this does
not really explain why the connections are stuck in CLOSE_WAIT because in that
case they wouldn't get closed at all.

-- 
You are receiving this mail because:
You are the assignee for the bug.