Re: exited on signal 11

[proforg <proforg@xxxxxxxxxxxx>]

да, строчки действительно есть, для всех падений кроме первого :) 

вот то что сказало gdb

(gdb)
#0  0xffffe410 in __kernel_vsyscall ()
#1  0xb7bb0691 in raise () from /lib/tls/i686/cmov/libc.so.6
#2  0xb7bb1f5b in abort () from /lib/tls/i686/cmov/libc.so.6
#3  0x080916f5 in ngx_worker_process_exit.133 (cycle=0x916f500) at ngx_process_cycle.c:952
#4  0x08053ec8 in ngx_worker_process_cycle.133 (cycle=0x0, data="" at ngx_process_cycle.c:733
#5  0x0804ded5 in ngx_spawn_process (cycle=0x0, proc=0x6, data="" name=0x83f42a0 "HM?\b\200B?\bдL?\bдL?\b", respawn=0) at ngx_process.c:187
#6  0x0804d4bd in ngx_master_process_cycle (cycle=0x0) at ngx_process_cycle.c:219
#7  0x0804a9f1 in main (argc=138366148, argv=0x83f4cc4) at nginx.c:295
(gdb) fr 3
#3  0x080916f5 in ngx_worker_process_exit.133 (cycle=0x916f500) at ngx_process_cycle.c:952
952     in ngx_process_cycle.c
(gdb) p c[i]
No symbol "i" in current context.
(gdb) p *c[i].read
No symbol "i" in current context.
(gdb) p *c[i].write
No symbol "i" in current context.
(gdb) p c
$1 = (ngx_connection_t *) 0x80d3740
(gdb) p *c
$2 = {data = "" read = 0x8379118, write = 0x817f548, fd = 105, recv = 0, send = 0, recv_chain = 0, send_chain = 0, listening = 0x83f4cdc, sent = 0,
  ctx = 0x83f50b0, servers = 0x86ee718, log = 0x83f4d18, pool = 0x0, sockaddr = 0x0, socklen = 0, addr_text = {len = 0, data = "" buffer = 0x0,
  number = 0, log_error = 0, buffered = 0, single_connection = 0, unexpected_eof = 0, timedout = 0, closed = 0, sendfile = 0, sndlowat = 0,
  tcp_nodelay = 0, tcp_nopush = 0}

И - спасибо за правку в ssi !


С уважением, 
Алексей Бещёков

On 12/6/05, Igor Sysoev < is@xxxxxxxxxxxxx> wrote:

On Tue, 6 Dec 2005, proforg wrote:

> хм, у меня в логах по утрам  пишет
> 2005/12/05 08:18:46 [notice] 22005#0: signal 17 (SIGCHLD) received
> 2005/12/05 08:18:46 [alert] 22005#0: worker process 11598 exited on signal 6
> (core dumped)
> 2005/12/05 08:18:46 [notice] 22005#0: signal 29 (SIGIO) received

> (gdb) bt
>#0  0xffffe410 in __kernel_vsyscall ()
>#1  0xb7bb0691 in raise () from /lib/tls/i686/cmov/libc.so.6
>#2  0xb7bb1f5b in abort () from /lib/tls/i686/cmov/libc.so.6
>#3  0x080916f5 in ngx_worker_process_exit.133 (cycle=0x916f500) at
>ngx_process_cycle.c:952
>#4  0x08053ec8 in ngx_worker_process_cycle.133 (cycle=0x0, data="">>at ngx_process_cycle.c:733
>#5  0x0804ded5 in ngx_spawn_process (cycle=0x0, proc=0x6, data="">>name=0x83f42a0 "HM?\b\200B?\bдL?\bдL?\b", respawn=0) at ngx_process.c:187
>#6  0x0804d4bd in ngx_master_process_cycle (cycle=0x0) at
>ngx_process_cycle.c:219
>#7  0x0804a9f1 in main (argc=138366148, argv=0x83f4cc4) at nginx.c:295

fr 3
p c[i]
p *c[i].read
p *c[i].write

>хотя может это и нормально :)

Это нормальная корка: по выходу nginx проверяет, не осталось ли открытых
соединений и если таковое нашлось, то вызывает abort(). Перед этим есть
что-нибудь про "open socket #NN left in NNN connection, aborting" ?


Игорь Сысоев
http://sysoev.ru