On Thu, 2009-05-21 at 22:39 +0100, sbr(at)rlfans.com wrote:
> Hi,
> 
> I noticed pound's memory usage slowly increasing over time so I've been
> having a quick scan through the code. I've noticed the following section of
> code (lines 883-893 of http.c) which I think is missing a call to
> free_headers:
> 
> /* add header if required */
> if(lstn->add_head != NULL)
>     if(BIO_printf(be, "%s\r\n", lstn->add_head) <= 0) {
>         str_be(buf, MAXBUF - 1, cur_backend);
>         end_req = cur_time();
>         logmsg(LOG_WARNING, "(%lx) e500 error write HTTPSHeader to %s: %s
> (%.3f sec)",
>         pthread_self(), buf, strerror(errno), (end_req - start_req) /
> 1000000.0);
>         err_reply(cl, h500, lstn->err500);
>         clean_all();
>         pthread_exit(NULL);
>     }
> 
> All the other pthread_exit calls before this follow a call to free_headers
> and free_headers is called just after this. Unfortunately as I don't use the
> add header option this isn't the cause of the memory problem I'm seeing.

Thanks Andrew, this goes into the next release.
-- 
Robert Segall
Apsis GmbH
Postfach, Uetikon am See, CH-8707
Tel: +41-44-920 4904

On Mon, 2009-05-18 at 19:52 +0200, Eirik Øverby wrote:
> Hi,
> 
> I keep seeing 'Line too long' messages in my logs; am wondering
>   - are these error mesages (i.e. is the request being served normally  
> despite the message?)
>   - why is MAXBUF lower than in 2.3?
> 
> I've recompiled with modified pound.h now; will there be adverse  
> effects, and if not, could this be modified in the next drop? The  
> lines in question are certificate information lines (subject, etc.),  
> and as I'm communicating exclusively with trusted parties on that  
> particular instance, I know there is nothing 'fishy' going on. I'm  
> also assuming requests ARE being served, as the client has not  
> complained yet..

In the newest version the default MAXBUF is 4k. You have a compile-time
option to make it larger if you wish.
-- 
Robert Segall
Apsis GmbH
Postfach, Uetikon am See, CH-8707
Tel: +41-44-920 4904

it does. this is great. yes, we have rather singular minded devs tho. I
will show this and the other response to them. thank u. we are balancing
two web servers. appreciate the help. I have just read through a good
portion of the o'reilly tomcat book and completely agree that there is no
reason to involve apache.

tim

On Thu, 18 Jun 2009 13:40:36 +0100, John Snowdon
<J.P.Snowdon(at)newcastle.ac.uk> wrote:
> Any reason why you list the service alpha.mysite.com and the same URL
> twice?
> 
> If you have more than one backend for the same URL then you should
> define it within the same service.
> 
> e.g.
> 
> 
> ListenHTTP
>         Address 192.168.1.42
>         Port    80
>         Service
>                 HeadRequire "Host:.*alpha.mysite.com.*"
>                 Url         "/index.*"
>                 BackEnd
>                         Address 192.168.1.5
>                         Port 80
>                 End
>                 BackEnd
> 		        Address 192.168.1.6
>                         Port 80
>                 End
>                 Session
>                         Type Basic
>                         TTL  300
>                 End
>         End
> End
> 
> I also agree with the last comment about making the connection direct to
> Tomcat, rather than through Apache and mod_jk if you are not using any of
> the more advanced Apache features (url rewriting and access control etc.)
> 
> Hope that helps
> 
> -John
> 
> On Wed, 2009-06-10 at 20:53 +0100, Tim Dunphy wrote:
>> ah, sorry! I was only including what I thought relevant. here is more
>> complete info. I will be looking into the manpage as suggested. but the
>> answer may not be obvious to me from the manpage alone, but I will do my
>> best.
>> 
>> 
>> thanks
>> 
>> ## Minimal sample pound.cfg
>> ##
>> ## see pound(8) for details
>> 
>> 
>> ######################################################################
>> ## global options:
>> 
>> User            "www-data"
>> Group           "www-data"
>> #RootJail       "/chroot/pound"
>> 
>> ## Logging: (goes to syslog by default)
>> ##      0       no logging
>> ##      1       normal
>> ##      2       extended
>> ##      3       Apache-style (common log format)
>> #LogLevel        1
>> 
>> ## check backend every X secs:
>> Alive           2
>> 
>> ## use hardware-accelleration card supported by openssl(1):
>> "pound.cfg.tomcat" 59L, 1337C
>> ## use hardware-accelleration card supported by openssl(1):
>> #SSLEngine      "<hw>"
>> 
>> 
>> ######################################################################
>> ## listen, redirect and ... to:
>> 
>> ## redirect all requests on port 80 ("ListenHTTP") to the virtual IP
>> address:
>> ListenHTTP
>>         Address 192.168.1.42
>>         Port    80
>> 
>>         Service
>>                 HeadRequire "Host:.*alpha.mysite.com.*"
>>                 Url         "/index.*"
>>                 BackEnd
>>                         Address 192.168.1.5
>>                         Port    80
>> 
>>                        Port    80
>>                 End
>>                 Session
>>                         Type Basic
>>                         TTL  300
>>                 End
>>         Service
>>                 HeadRequire "Host:.*alpha.mysite.com.*"
>>                 Url         "/index.*"
>>                 BackEnd
>>             Address 192.168.1.6
>>                         Port 80
>>                 End
>>                 Session
>>                         Type Basic
>>                         TTL  300
>>                 END
>>         End
>> End
>> 
>> 
>> desired url:
>> 
>> http://alpha.mysite.com/index.jsf
>> 
>> error:
> 
> 
> --
> To unsubscribe send an email with subject unsubscribe to pound(at)apsis.ch.
> Please contact roseg(at)apsis.ch for questions.

-- 
"You were right about the stars. Each one is a setting sun." - Wilco

Yes unless you need hybrid features from apache you usually do not
need apache in front of tomcat. Also I once benchmarked mod_jk and as
you can suspect there is overhead. I do not have the numbers any more,
but it makes sense that you have apache overhead, mod_jk overhead, and
then finally tomcat overhead. It is fairly significant.

On Thu, Jun 18, 2009 at 10:33 PM, Tim Dunphy<bluethundr(at)nylsd.com> wrote:
>
> it does. this is great. yes, we have rather singular minded devs tho. I
> will show this and the other response to them. thank u. we are balancing
> two web servers. appreciate the help. I have just read through a good
> portion of the o'reilly tomcat book and completely agree that there is no
> reason to involve apache.
>
> tim
>
> On Thu, 18 Jun 2009 13:40:36 +0100, John Snowdon
> <J.P.Snowdon(at)newcastle.ac.uk> wrote:
>> Any reason why you list the service alpha.mysite.com and the same URL
>> twice?
>>
>> If you have more than one backend for the same URL then you should
>> define it within the same service.
>>
>> e.g.
>>
>>
>> ListenHTTP
>>         Address 192.168.1.42
>>         Port    80
>>         Service
>>                 HeadRequire "Host:.*alpha.mysite.com.*"
>>                 Url         "/index.*"
>>                 BackEnd
>>                         Address 192.168.1.5
>>                         Port 80
>>                 End
>>                 BackEnd
>>                       Address 192.168.1.6
>>                         Port 80
>>                 End
>>                 Session
>>                         Type Basic
>>                         TTL  300
>>                 End
>>         End
>> End
>>
>> I also agree with the last comment about making the connection direct to
>> Tomcat, rather than through Apache and mod_jk if you are not using any of
>> the more advanced Apache features (url rewriting and access control etc.)
>>
>> Hope that helps
>>
>> -John
>>
>> On Wed, 2009-06-10 at 20:53 +0100, Tim Dunphy wrote:
>>> ah, sorry! I was only including what I thought relevant. here is more
>>> complete info. I will be looking into the manpage as suggested. but the
>>> answer may not be obvious to me from the manpage alone, but I will do my
>>> best.
>>>
>>>
>>> thanks
>>>
>>> ## Minimal sample pound.cfg
>>> ##
>>> ## see pound(8) for details
>>>
>>>
>>> ######################################################################
>>> ## global options:
>>>
>>> User            "www-data"
>>> Group           "www-data"
>>> #RootJail       "/chroot/pound"
>>>
>>> ## Logging: (goes to syslog by default)
>>> ##      0       no logging
>>> ##      1       normal
>>> ##      2       extended
>>> ##      3       Apache-style (common log format)
>>> #LogLevel        1
>>>
>>> ## check backend every X secs:
>>> Alive           2
>>>
>>> ## use hardware-accelleration card supported by openssl(1):
>>> "pound.cfg.tomcat" 59L, 1337C
>>> ## use hardware-accelleration card supported by openssl(1):
>>> #SSLEngine      "<hw>"
>>>
>>>
>>> ######################################################################
>>> ## listen, redirect and ... to:
>>>
>>> ## redirect all requests on port 80 ("ListenHTTP") to the virtual IP
>>> address:
>>> ListenHTTP
>>>         Address 192.168.1.42
>>>         Port    80
>>>
>>>         Service
>>>                 HeadRequire "Host:.*alpha.mysite.com.*"
>>>                 Url         "/index.*"
>>>                 BackEnd
>>>                         Address 192.168.1.5
>>>                         Port    80
>>>
>>>                        Port    80
>>>                 End
>>>                 Session
>>>                         Type Basic
>>>                         TTL  300
>>>                 End
>>>         Service
>>>                 HeadRequire "Host:.*alpha.mysite.com.*"
>>>                 Url         "/index.*"
>>>                 BackEnd
>>>             Address 192.168.1.6
>>>                         Port 80
>>>                 End
>>>                 Session
>>>                         Type Basic
>>>                         TTL  300
>>>                 END
>>>         End
>>> End
>>>
>>>
>>> desired url:
>>>
>>> http://alpha.mysite.com/index.jsf
>>>
>>> error:
>>
>>
>> --
>> To unsubscribe send an email with subject unsubscribe to pound(at)apsis.ch.
>> Please contact roseg(at)apsis.ch for questions.
>
> --
> "You were right about the stars. Each one is a setting sun." - Wilco
>
> --
> To unsubscribe send an email with subject unsubscribe to pound(at)apsis.ch.
> Please contact roseg(at)apsis.ch for questions.
>