I need to direct traffic to different servers based solely on the domain 
name part of the url (which, if I understand things, is not technically 
a part of the url).

In other words, I need all HTTP requests for projects.greywether.com to 
go to one server, and all HTTP requests for www.expeditiontea.com to go 
to another server.  The are both domains whose server names ("projects" 
and "www") point to the same IP address that pound is listening on.

I looked at doing this with a UrlGroup, that seemed like the natural 
thing to do.  But then I discovered that the UrlGroup only has the 
non-domain-name portion of the Url to match against.

I do know about HeadRequire Host for matching against the domain name, 
but that happens within a UrlGroup, right?  In other words, I can't use 
it to fork requests between different servers, right?  Or am I 
misunderstanding something about how UrlGroups work?

Suggestions or corrections are much appreciated.

Thanks!

- Gary

P.S.  And a bleated thanks to Robert for answering my last question 
about HTTP/SSL.  For some reason I didn't get subscribed correctly the 
first time around and didn't see the response until today when I 
searched the archives.  Looks like I'm subscribed correctly now.  Thanks 
again.

Gary Affonso wrote:[...]
               ^^^^^^^

Uh that would be belated thanks.  Damn spell check let that one by.  I'm 
mortified. :-)

- Gary

On Wed, 13 Jul 2005 21:50:38 -0700 Gary Affonso
<glists(at)greywether.com>
wrote:
[...]

Yes, you do use the UrlGroup, just matching everything.

UrlGroup ".*"
HeadRequire Host ".*projects.greywether.com.*"
BackEnd xxx.xxx.xxx.xxx,80,1
EndGroup

UrlGroup ".*"
HeadRequire Host ".*www.expeditiontea.com.*"
BackEnd yyy.yyy.yyy.yyy,80,1
EndGroup
[...]

Some spelling errors are more embarrassing than others...[...]

Robert Segall wrote:[...]

Ahhhh.

Is it just me, or is the documentation (the man page) rather unclear on 
this point?

What I get when I read the man-page is that matching is done against the 
first URLGroup filter and then *after* that match, within the URLGroup 
on things like HeadRequire Host.

Meaning that the above example would not work because the first .* would 
match and the other .* filters (each with different HeadRequire 
directives) would never be seen.

Looks like, though, Pound matches on a combination of the UrlGroup 
filter + the HeadRequire filter.  Almost as if the HeadRequire filter 
was just an extension of the UrlGroup filter.  So you can have many .* 
UrlGroup filters and they all still get seen.

I'm brand-new to Pound and am unclear how code/documentation 
contributions are handled, but I'd be happy to make some mods to the 
man-pages (I have professional technical writing experience) to make 
that point more clear (at least to me :-).

And speaking of docs, is there a reason a set of online (web) docs don't 
exist (even if it was just a copy of the man page)?  Sure is handy to 
have a web page, not just "man".
[...][...]

That one particularly so.  I still cringe when I see it.  Bahhhh. :-)

- Gary

On Fri, 15 Jul 2005 08:57:26 -0700 Gary Affonso
<glists(at)greywether.com>
wrote:
[...]

We welcome improvements and patches of code and documentation. Just drop
me a message and we'll try to integrate the stuff.
[...]

Pure laziness. We are not in the business of doing web sites - the Pound
page "just growed up".[...]