Two RaQs, both RaQ 4i both at rackshack, configured as similar as two servers can be who were setup almost 4 months apart.

Server 1 is "heavily loaded" Server 2 is not. Server 1 is not the problem, Server 2 is, as you can see if you load the images.

Server 1:
http://www.dixiesys.com/imagetest/

Server 2:
http://www2.dixiesys.com/imagetest/

These images were uploaded from the same computer, using BPFTP.

I've had 2 customers on the #2 server complain about corrupted image uploads. There are roughly 45 domains on that server.

I've only ran across one other thread about this and this is the error created in the log files:

Jan 5 11:55:51 www3 proftpd[18242]: 216.40.243.5 (ppp-09.XX***********[XXX.XXX.XXX.XXX]) - Data transfer stall timeout: 3600 seconds

However my uploads created no errors but the files are obviously fubared.

I uploaded another batch of files via SCP (all of them were 46K or so in size) and so far out of about 100 files I've seen no corruption at all (and this was uploaded to the server giving problems).

It seems like there might be like, a buffer that if it gets filled might be the problem but small files don't fill it, something like that, if anyone has a CLUE on this one feel free to chime in.

Oh I've uploaded 300-800K zip/gzip files about 12 so far, and NONE have gotten corrupted, so WHY images? why ONLY images (it seems only images anyway) have this problem? and why only on the one server?

A cookie to anyone with a clue on this one!

Sorry, I have to ask the obvious :( ...but some of them really look like ascii uploads. That would also explain the "images only" part.

Yep they do, but no they definitely weren't as I uploaded these myself as a test.

I compiled a newer version of Proftpd and installed it and this seems to have fixed it.