Robert,

 

I tried making the changes you suggested to help with downloading large
files, and I got it to work!!  This actually allowed me to work-around the
issue I was having with the server code generating mixed http 1.1 and 1.0
code and the problems I had with the client timeout values.

 

The key to fix my problem was to increase the tcp_wmem setting.  The
hardware I run Pound on only has 64M of memory, so it was starving for
memory from the get-go.  None of the other suggestions like changing http.c
of changing tcp_mem made a difference, but in my case tcp_wmem was the key.
I can now turn down the client timeout value to 1, and have PDF files get
served up quickly, and allow clients to download large attachments without
corruption.

 

I increased the minimum tcp_wmem setting to the largest file size I could
imagine downloading, and now all files download to IE without corruption.
With a lower setting I would either get a client timeout, or a server
timeout error.

 

Thanks for the great suggestions, I can finally move on to other problems to
solve!

 

Sincerely,

Steven

 

On Wed, 27 Jul 2005 15:54:23 -0500 "Steven Vlach" <Home_Boy(at)msn.com>
wrote:

> Robert,
> 
>  
> 
> I tried making the changes you suggested to help with downloading
> large files, and I got it to work!!  This actually allowed me to
> work-around the issue I was having with the server code generating
> mixed http 1.1 and 1.0 code and the problems I had with the client
> timeout values.
> 
>  
> 
> The key to fix my problem was to increase the tcp_wmem setting.  The
> hardware I run Pound on only has 64M of memory, so it was starving for
> memory from the get-go.  None of the other suggestions like changing
> http.c of changing tcp_mem made a difference, but in my case tcp_wmem
> was the key. I can now turn down the client timeout value to 1, and
> have PDF files get served up quickly, and allow clients to download
> large attachments without corruption.
> 
>  
> 
> I increased the minimum tcp_wmem setting to the largest file size I
> could imagine downloading, and now all files download to IE without
> corruption. With a lower setting I would either get a client timeout,
> or a server timeout error.
> 
>  
> 
> Thanks for the great suggestions, I can finally move on to other
> problems to solve!
> 
>  
> 
> Sincerely,
> 
> Steven

Glad to hear it works - good news are always welcome. Thanks for the
feedback. In any case we'll keep looking for a better solution than
this, but for the meantime I'm sure this is helpful.
-- 
Robert Segall
Apsis GmbH
Postfach, Uetikon am See, CH-8707
Tel: +41-44-920 4904