Hi all,
I have 4 web applications that need to be proxied. Three of these apps are in
the same box but running on different ports.
Assuming that pound.sample.com is the server running pound service,
server1.sample.com is the first server and server2.sample.com is the second
server.
What I'm trying to achieve is like this:
1. http://pound.sample.com ---> http://server1.sample.com:80
2. https://pound.sample.com/app1
---> http://server1.sample.com:9001
3. https://pound.sample.com ---> http://server1.sample.com:9002
4. https://pound.sample.com/app2
---> http://server2.sample.com:80

Can somene provide me a sample config on how pound works on this scenario?

Thanks.



[...]

On Sunday 05 December 2004 04.21, 5pit00n wrote:[...]

Pound can't do this - it would involve URL rewriting, which is not supported. 
It could do it without changing the actual path (hint: use the UrlGroup 
pattern matching capabilities).

As an aside: when you get into such a situation it may be a good time to 
rethink your application...[...]

Robert Segall wrote:[...]

   (jumping into this thread)

   I suppose it depends on how you look at things... but...

   If you approach everything from the point of view of pound, a common 
theme on this list seems to be that the fault lies with "badly designed 
applications".

   The reflex response to any configuration question or special request 
seems to be "can't do this because your application is badly designed".

   I have to argue that a lot of applications are designed the way they 
are for very good reasons. I would also argue that designing your system 
to best support your application rather than your ancillaries is not a 
bad thing.

   I don't really have a conclusion to this email, it's more of an 
observation coupled with a bit of frustration.

   I suspect that pound would get a lot more use by a lot more people if 
it was more adaptable to real-world problems, rather than seeming to 
require the application to be adapted to pound.
[...]

On Monday 06 December 2004 21.36, Trevor Boicey wrote:[...]

Thanks for the remarks Trevor

I certainly didn't wish to imply that the application is "bad" (whatever that 
may mean). However, based on my own experience, I know that when I get into 
such a situation, I usually start thinking about the "bigger picture", and I 
am sure you have also come across such "pathological cases". All in all they 
are a good thing, as they force one to reconsider the basic assumptions, or 
perhaps reflect on how the design got into such a state. I trust you would 
not call that a "reflex reaction", but rather a well-intentioned piece of 
advice. Like all free advice you are welcome to disregard it - that is most 
certainly your privilege.

If you read the original response you may notice that I said Pound can't do it 
(due to lack of URL rewriting), but it is perfectly doable if you want to put 
up with the full path from the request (to take the original example: not 
https://pound.sample.com/app1
---> http://server1.sample.com:9001 but
rather 
https://pound.sample.com/app1
---> http://server1.sample.com:9001/app1).
It 
may be the case that you think URL rewriting to be an important feature - and 
as always if enough people support your view this will be added to a future 
version. Contributing some code, even if only a skeleton or outline, would 
accelerate the process somewhat.

On the other hand, as the owner of the application it is my duty to try and 
resist the dreaded "creeping featuritis" so prevalent in today's programs. 
I'll thus try and ask some hard questions (playing devil's advocate to some 
extent) just to make sure the proposed extension makes sense. I hope you 
won't find that too offensive and that you'll continue contributing in the 
future - hopefully your goal, just like mine, is to have a better program.[...]

>    If you approach everything from the point of view of pound, a
common[...]

I have to disagree. First, this list is devoted to taking everything
from the point of view of pound. Then, if something can't be done with
pound, well, no silver bullet, it's just a tool, and a good one too.
[...]

Uhm, may be, but a lot are not like that. The original question
[...]

looks like there's just a bunch of separate web applications thrown
on top of each other, at least so it appears. It doesn't mean they
are ill designed of course, but why pound has to adapt to them now ?
[...]

I hear you, I wish myself there's a lot of additional features in
pound, but as soon as the author tries to keep it simple, then so be
it, hopefully it helps keeping is robust.

Sincerely,
Dmitry Dvoinikov
http://www.targeted.org/

--- Original message follows ---
[...][...]
[...]
[...]
[...]
[...]
[...]
[...]
[...]

Well, thanks anyway for all the replies. Guess I should stick with apache then
since well, it supports my badly designed apps. Why can't all you people just
go straight to the point whether it's a yes or no. There's no need to
cricticize about the way the apps were being designed.
In fact, you didn't know half of the story here which I'm not intend to share
with all of you.
I respect Robert's answers though, just the thing that I expected, straight to
the point and why it won't work with pound. 

Well, keep up the good work, Bob. To all the trolls, show off your mad skills
somewhere else.

Pissed,
5pit00n


 --- On Mon 12/06, Dmitry Dvoinikov < dmitry(at)targeted.org > wrote:
From: Dmitry Dvoinikov [mailto: dmitry(at)targeted.org]
To: tboicey(at)brit.ca
     Cc: pound(at)apsis.ch
Date: Tue, 7 Dec 2004 09:30:24 +0500
Subject: Re[2]: newbie config help...
[...]
[...]

5pit00n wrote:[...]

   I feel I should reply again, since I seem to have stirred the pot a 
little bit.

   Like a lot of users on this list, I am not actually a pound user... 
yet. I am "looking into" pound to solve a problem, in my case, mirroring 
and failover.

   The application is designed, written, and has been running happily 
for a while. Now I want to make sure it stays running, so am looking to 
beef up the architecture.

   This isn't a rare case though, in the real world it's pretty much the 
way it is. You start with an application and it grows into something you 
care about, and then you start to worry about infrastructure. Or, you 
have an app, and have problems with your infrastructure, so you are 
looking for alternatives.

   Designing an app and the infrastructure at the same time from the 
ground up sounds like a "great plan", but it's just not the way a lot of 
projects go.

   So... a lot of people like myself are looking at pound, and have an 
existing application design. Most of us are generally happy with our 
application design as well, and believe it is the way it is for good 
reasons.

   (quick summary of my situation, I have an app and want a hot-swap 
mirror setup. The app does a LOT of caching to handle complicated 
processing and displaying of results quickly, so sharing the load 
between two servers is very problematic as data writes on the server 
must intelligently decide which caches are dirty, perform a lot of 
calculations, and then get ready to serve the new data... to load 
balance this is problematic)

   However, since I do not want any traffic going to the mirror at all 
unless the primary is dead, it's not supported under pound, and was told 
somewhat that "my application design was bad". Paraphrasing of course.

   I'm not whining, I have no right to assume that the software will 
solve any of my problems, that's the way it is.

   However, I do feel that most people who write software, especially 
open source software, want to see it used and help people. Understanding 
that pound is most likely to be a tool that has to adapt to a situation 
rather than the other way around might be a step that has to be taken to 
appeal to a wider audience.
[...]