On Fri, 2006-09-01 at 15:47 +0200, Maciej Bogucki wrote:[...]

I've run into this in apache - it's probably FD_SETSIZE, which usually
defaults to 1024.  I've found that I have to raise that and recompile as
well as using ulimit.  It's defined in /usr/include/bits/typesizes.h on
my system.
[...]

On Fri, 2006-09-01 at 14:19 +0200, dirk dekker wrote:[...]

You'll have to ask the Ubuntu people - we have nothing to do with it.[...]

On Fri, 2006-09-01 at 10:05 -0600, Sean Gabriel Heacock wrote:[...]

Sorry, but this is very unlikely. Pound has moved from select() to
poll() a long time ago precisely in order to avoid these problems.

As to the OP: I don't really know, but I suspect you start Pound with
the wrong ulimit. Why don't you try printing out the ulimit from within
Pound itself and see what it says (or at least use ulimit in your start
script for an indication)?[...]

Hello Robert ,
Thanks!
I have forwarde it now to : nobse(at)debian.org
Dirk

-----Oorspronkelijk bericht-----
Van: Robert Segall [mailto:roseg(at)apsis.ch] 
Verzonden: vrijdag 1 september 2006 18:15
Aan: pound(at)apsis.ch
Onderwerp: Re: [Pound Mailing List] Ubuntu packages i386

On Fri, 2006-09-01 at 14:19 +0200, dirk dekker wrote:[...]

You'll have to ask the Ubuntu people - we have nothing to do with it.
--
Robert Segall
Apsis GmbH
Postfach, Uetikon am See, CH-8707
Tel: +41-44-920 4904

[...]

Regardless of this limitation on the number of sockets, you will probably
run into problems before long with loading.

You may need to investigate TCP level load balancing between multiple
instances of pound on separate machines.  In preparation for this, you may
want to look at making sure that you can do without sticky sessions.  IP
based load balancing may suit your needs and it is much more likely to work
with multiple pounds in front of your applications.


On 9/1/06 6:47 AM, "Maciej Bogucki" <maciej.bogucki(at)artegence.com>
wrote:
[...][...][...][...][...]

Return Receipt
                                                                           
   Your       RE: [Pound Mailing List] Ubuntu packages i386                
   document:                                                               
                                                                           
   was        jbrown(at)camsys.com                                            
   received                                                                
   by:                                                                     
                                                                           
   at:        09/01/2006 01:47:04 PM

On Wed, 2006-09-06 at 10:02 +0200, Olivier Rossel wrote:[...]

Yes. See the ClientCert option.[...]

Quoting aT <pakistan(at)gmail.com>:
[...]

linux-ha stuff would be a solution I guess

we just have a redundant server on hot stand by I can "flip on" by  
running a shell script if the first one goes down. we have nagios  
monitoring server health. not very high tech.
[...]

Hi,

actualy we're running two FreeBSD-Boxes with carp configured fpr
fail-over. On both machines pound is running with an equal configuration.

In the near future we will migrate to OpenBSD because of some strange
behaviors with the carp-implementation on FreeBSD. I would like to see a
failover-mechanism with HA-Beat and session takeover, but the actual
setup is better than nothing :)

Regards,

Falk Brockerhoff

We use heartbeat (www.ultramonkey.org)  to provide a virtual IP for pound to
run on two separate servers.


-----Original Message-----
From: aT [mailto:pakistan(at)gmail.com] 
Sent: Monday, 11 September 2006 12:01 AM
To: pound(at)apsis.ch
Subject: [Pound Mailing List] Pound failover

Hi ,

Pound provides  great fail over and load balancing aspect to web sites.
But how people are making Pound redundant ?
I mean if i have a single pound server serving requests and this one fails ,
is there a way of automatically shifting load to another pound ?
is DNS round robin  the only solution ?

Regards


--
Syed Atif Ali
D. +971 4 3911914
F. +971 4 3911915[...]

I'd be interested to hear further information on this setup - it's
something I will be attempting shortly to move away from the 'pound as
single point of failure' situation at present.

How do you manage your pound configuration and restarts across the
multiple servers?

 John Snowdon - IT Support Specialist
-==========================================-
 School of Medical Education Development 
 Faculty of Medical Sciences Computing
 University of Newcastle

[...]
26758000/1157954219000

Hello,

aT wrote:[...]
[...]

Pound 2.1 on OpenBSD 3.9 with pre-empted CARP, works well for me.

aT wrote:[...]

Our Pound config rarely changes, so it's a manual thing, Im sure you could
use rsync if necessary, or if your config changes alot.
The heartbeat software manages the virtual IP and also the pound stop/starts
if a server fails.  It's a fairly simple configuration, but we haven't had
any issues with it. 

-----Original Message-----
From: John Snowdon [mailto:J.P.Snowdon(at)newcastle.ac.uk] 
Sent: Monday, 11 September 2006 8:02 PM
To: pound(at)apsis.ch
Subject: RE: [Pound Mailing List] Pound failover

I'd be interested to hear further information on this setup - it's something
I will be attempting shortly to move away from the 'pound as single point of
failure' situation at present.

How do you manage your pound configuration and restarts across the multiple
servers?

 John Snowdon - IT Support Specialist
-==============-
 School of Medical Education Development  Faculty of Medical Sciences
Computing  University of Newcastle

[...]
26758000/1157954219000
[...]

Hi Tim.. it's a regular expression (man pound); 

HeadRequire     "Host:.*domain.com.*" 

...should get what you want.

 John Snowdon - IT Support Specialist
-==========================================-
 School of Medical Education Development 
 Faculty of Medical Sciences Computing
 University of Newcastle

 Email : j.p.snowdon(at)ncl.ac.uk

[...]

The pattern matching is regular expression based. I would suggest boning up 
on regular expressions.

Regards
Richard


----- Original Message ----- 
From: "Tim Perrett" <tperrett(at)butlerandtanner.com>
To: <pound(at)apsis.ch>
Sent: Tuesday, September 19, 2006 11:44 AM
Subject: [Pound Mailing List] Www matching

[...]

Hi,

I think that svn:// protocol is not http protocol... so you can't
use pound for this. But maybe I'm wrong.

If you want pound in front of subversion server, I think you have to
use apache subversion server and http protocol.

--
 .''`. Ondra 'Kepi' Kudlik
: :' : Debian GNU/Linux User
`. `'
  `-   http://www.nosoftwarepatents.com/cz/m/intro/index.html



Pá, zář 22, 2006 ve 09:47:17 +0100, Tim Perrett napsal:[...]

huh? pound is a HTTP proxy, but svn: is not a HTTP-based protocol.

To access subversion through pound, you will have to use WebDAV. I have an
apache intermediary configured behind pound that does the subversion
interfacing.

-C

Zitat von Tim Perrett <tperrett(at)butlerandtanner.com>:
[...]

You should be able to run svn and pound side by side. What you need to
understand is pound is only an HTTP proxy. Just take out all references
to svn and port 8200 from pound's config, restart it, and make sure that
svn is up and listening to 8200 and you should be good. Or even better
yet, access svn through ssh as I am. Instructions are in the svn manual
online. If you need to proxy the svn traffic off to some other machine
though... that's a different can of worms.

Tim Perrett wrote:[...]

Put svn behind a web-dav capable HTTP server like Apache (details are in the
SVN book) and you will be able to route all of your svn traffic through
https URL's that pound will secure for you.


On 9/22/06 8:28 AM, "Jon Garvin" <jgarvin.lists(at)gmail.com> wrote:
[...][...][...]