[Bug 212920] Li loaded web server cath race condition on _close () from /lib/libc.so.7 with accf_http

[bugzilla-noreply at freebsd.org]

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=212920

Mikanoshi <iam at mikanoshi.name> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |iam at mikanoshi.name

--- Comment #1 from Mikanoshi <iam at mikanoshi.name> ---
I have the same problem on 11.0-RC3 if reset_timedout_connection is enabled in
Nginx (accept filters are enabled too).

Running truss on a worker stuck in soclos produces this output:
    close(49) ERR#9 'Bad file descriptor'
    write(36,"2016/09/19 18:16:03 [crit] 15495#0: close() socket 49 failed (9:
Bad file descriptor)\n",86) = 86 (0x56)
and then worker starts to process requests as usual.

Such entries are written to log when reset_timedout_connection is disabled
(workers do not stuck in this case):
[info] 89888#0: close() socket 12 failed (54: Connection reset by peer)

Problem appeared after moving from FreeBSD 9.2 to 11.0-RC3.

Discussion on Nginx forum (in russian):
https://forum.nginx.org/read.php?21,269501,269778#msg-269778

-- 
You are receiving this mail because:
You are on the CC list for the bug.