Hello,

  Pound does not want to shutdown on freebsd 4.9.
  I installed it using patches from freebsd ports.
  In detail:
     CFLAGS=-DF_CONF=\"$(F_CONF)\"   -O   -pipe   -DAEMON   -D  UNSAFE
     -D_THREAD_SAFE -DNEED_STACK
     LIBS=-lssl -lcrypto

  What happens is. When I try to shutdown pound it quits accepting new
  connections  (it  seems  so),  but  it  does  not  want  to shutdown
  completely.  'top' shows 'soclos' in status. 'netstat' shown several
  connections established.

  What can you suggest?

  P.S.  New pound does not want to start, because it cant bind to port
  80.

-- 
/ Pavel Merdine

On Tuesday 09 December 2003 18:49, Pavel Merdine wrote:
> Hello,
>
>   Pound does not want to shutdown on freebsd 4.9.
>   I installed it using patches from freebsd ports.
>   In detail:
>      CFLAGS=-DF_CONF=\"$(F_CONF)\"   -O   -pipe   -DAEMON   -D  UNSAFE
>      -D_THREAD_SAFE -DNEED_STACK
>      LIBS=-lssl -lcrypto
>
>   What happens is. When I try to shutdown pound it quits accepting new
>   connections  (it  seems  so),  but  it  does  not  want  to shutdown
>   completely.  'top' shows 'soclos' in status. 'netstat' shown several
>   connections established.
>
>   What can you suggest?
>
>   P.S.  New pound does not want to start, because it cant bind to port
>   80.

What does "shut down" mean to you? Is there anything in the logs? More info 
required here...
-- 
Robert Segall
Apsis GmbH
Postfach, Uetikon am See, CH-8707
Tel: +41-1-920 4904

On Wednesday 10 December 2003 13:34, Pavel Merdine wrote:
> > What does "shut down" mean to you?
>
> I mean "killall pound"
>
> > Is there anything in the logs? More info required here...
>
> Dec 10 11:58:06 balancer pound: received signal 15 - exiting...
>
> that's all.
>
> the interesting thing is sometimes is starts after stopping.
> but I see two processes in top:
>
> 534 www        2   0 24520K 10048K poll     0:08  7.75%  7.18% pound
> 375 www        2   0 88084K 74984K soclos 120:05  0.00%  0.00% pound
>
> This  case  is fine for me. I'll try to catch logs when pound does not
> start again.

So you say Pound does not exit after receiving a signal 15? Try 9 ;-)

It looks like you get stuck zombie-like on some blocking socket I/O. Given 
that this does not happen on any other system (including FreeeBSD in earlier 
versions) it may have something to do with 4.9. Could you try and pinpoint 
the problem?

Secondly, I find it really strange that you would have more than one Pound 
process. The *BSDs implement threads as a user-space library, thus all 
threads run in a single process. Could it be that you started multiple 
instances?

Also: please keep this on the list.
-- 
Robert Segall
Apsis GmbH
Postfach, Uetikon am See, CH-8707
Tel: +41-1-920 4904

Hello ,

Wednesday, December 10, 2003, 4:43:40 PM, you wrote:

>> > Is there anything in the logs? More info required here...
>>
>> Dec 10 11:58:06 balancer pound: received signal 15 - exiting...
>>
>> that's all.
>>
>> the interesting thing is sometimes is starts after stopping.
>> but I see two processes in top:
>>
>> 534 www        2   0 24520K 10048K poll     0:08  7.75%  7.18% pound
>> 375 www        2   0 88084K 74984K soclos 120:05  0.00%  0.00% pound
>>
>> This  case  is fine for me. I'll try to catch logs when pound does not
>> start again.

> So you say Pound does not exit after receiving a signal 15? Try 9 ;-)

I forgot to mention. It cannot be killed even with kill -9 :)
So maybe you're right about FreeBSD.

> It looks like you get stuck zombie-like on some blocking socket I/O. Given 
> that this does not happen on any other system (including FreeeBSD in earlier 
> versions) it may have something to do with 4.9. Could you try and pinpoint 
> the problem?

The  problem is that this happens only in production environment (when
there  is  lots  of  connections).  I'll try to do something when time
allows.

> Secondly, I find it really strange that you would have more than one Pound 
> process. The *BSDs implement threads as a user-space library, thus all 
> threads run in a single process. Could it be that you started multiple 
> instances?

The process 375 you see is zombee one. It disappeared after some time.

> Also: please keep this on the list.

Sorry,  I  forgot to tune my mail client for this list to do it for me
:)

-- 
/ Pavel Merdine