Re: Downloads to Win2K corrupted behind Linux ADSL router

[followup-to set]
In article <Xns93CAA5CC6F493jpmvrealtime@216.128.74.129>,
Martin Vuille wrote:
> Here's the problem: if I try to download a file from the Internet
> (via HTTP) to either of the Win2K boxes, the transfers completes
> correctly, the tranferred file has the right length, but the data
> in the file is corrupt.
>
> But, if I download the very same file (also via HTTP) to the Linux
> box, the data in the file is OK. 100% of the time.

Your Linux box should have a program "md5sum" installed. Use that to
compare the files. ("man md5sum" ... it's very easy to use.)

> How could the other end possibly know the difference between
> the Linux box and the Win2K boxes?

This one has an easy and obvious answer: the HTTP client sends an
identification string to the server upon request. Microsoft (among
others) has been known to use this feature to block out non-approved
browsers.

It could be this, although that seems unlikely. It's also possible that
the HTTP client request being sent is malformed. So try different
clients on Windows ... Opera? Netscape? Try wget for Windows. You could
also try Cygwin, and use any of the POSIX-compliant browsers available.

You didn't mention if this is with just one server, or with multiple
sites. The URL alone might be misleading: it's easy and commonplace to
host many domains on a single server. Try from a server you KNOW to be
unrelated to the ones you've already tried.

Do you have a Linux system behind the firewall? Does it have similar
problems? I wouldn't think that the NAT/packet forwarding process could
cause this kind of issue, but it might not hurt to rule that out. First
try the different clients and different servers; that seems more likely
to be the source of the problem.

The Windows machines are trying to download from a remote site to their
own local filesystem, correct? It's not saving the file to a Samba or
other network fileshare, is it? If it's a network share, the problem
could be with the server providing the share.
--
/dev/rob0 - preferred_email=i$((28*28+28))@softhome.net
or put "not-spam" or "/dev/rob0" in Subject header to reply

/dev/rob0 <rob0@gmx.co.uk> wrote in news:slrnbill9q.iku.rob0
@linuxbox.stpaultel.com:

> [followup-to set]
> In article <Xns93CAA5CC6F493jpmvrealtime@216.128.74.129>,
> Martin Vuille wrote:
>> Here's the problem: if I try to download a file from the Internet
>> (via HTTP) to either of the Win2K boxes, the transfers completes
>> correctly, the tranferred file has the right length, but the data
>> in the file is corrupt.
>>
>> But, if I download the very same file (also via HTTP) to the Linux
>> box, the data in the file is OK. 100% of the time.
>
> Your Linux box should have a program "md5sum" installed. Use that to
> compare the files. ("man md5sum" ... it's very easy to use.)

Sorry, but I don't understand your point. I already know that the
files are received differently by Linux and Win2K, based on a
simple binary compare.

>
>> How could the other end possibly know the difference between
>> the Linux box and the Win2K boxes?
>
> This one has an easy and obvious answer: the HTTP client sends an
> identification string to the server upon request. Microsoft (among
> others) has been known to use this feature to block out non-approved
> browsers.
>

I thought of that, so I tried the download using wget on both Linux
and Win2K and checked that the identification string was the same
in both cases. The results were the same: Linux OK, Win2K corrupt.

> You didn't mention if this is with just one server, or with multiple
> sites. The URL alone might be misleading: it's easy and commonplace to
> host many domains on a single server. Try from a server you KNOW to be
> unrelated to the ones you've already tried.

I originally ran into this problem while trying to download patches
from the Microsoft web site (which turns out to be an ms.akamai.net
site.) So my "test files" were all from the Microsoft web site.

I just tried again now, and 4/5 of the files I tried to d/l from the
Microsoft web site came through OK, and 1/5 was corrupt. Last week,
the files were corrupt 100% of the time. The one that was corrupt
today was also corrupt last week. The other ones are new files that
I had not tried to d/l last week.

I also tried downloading some files from Tucows, and all the files
downloaded cleanly.

On the other hand, I was and still am seeing occasional corruption
in POP3 traffic. I.e., there will be garbage characters in the e-mails
I retrieve from my ISP, or attachments can't be retrieved cleanly.


> The Windows machines are trying to download from a remote site to their
> own local filesystem, correct? It's not saving the file to a Samba or
> other network fileshare, is it? If it's a network share, the problem
> could be with the server providing the share.

Yes, the Win2K boxes are d/l'ing to their own file systems.

MV

--
Do not send e-mail to the above address. I do not read e-mail sent there.