Web Hosting Talk







View Full Version : Non-Forking Web Servers that support resume?


Groo
04-05-2001, 01:29 PM
I've tried:

thttp - supports resume but takes up TOO MUCH memory (go figure)

webfs - awesome, amazing webserver to just serve static files/images - totally barebones, NO RESUME

aolserver - lots of pros..seems ok..but.. NO RESUME

boa - better than webfs, but again, no resume

I server a lot of big files (mp3, mpegs, etc.) and the ability to support resume (RANGE: xxxxxxx- in http headers) is necessary... Apache does this but apache is too much of a heavyweight for this...

Anyone have any suggestions as to what else I should try? To run on FreeBSD...

Thanks

-- on a side note, I do not have root on this machine, it's a 'virtual server' so I have a partition allocated to me type of deal.. So I can't go into /usr/libexec and install libraries, some web servers I've gone out to try (mathopl, xitami) need to get libc.so.5 from there and it's not and I can't find a place to change it in the sources.. maybe i'm blind.

Groo
06-13-2001, 09:43 PM
Wow somehow I came across this thread again :D

In case any of you care / were curious about the result of this...

WebFS correctly supports resume now...

About 3 days after this post, I worked with the author to fix byte-range offsetting as it wasn't http 1.1 compliant (incorrect variable initialization was screwing everything up).

You can get it at http://www.strusel007.de/linux/webfs.html

qslack
06-13-2001, 10:43 PM
The original poster really should use FTP.

HTTP isn't and wasn't meant for large files; FTP is *much* more efficient for 'em.

Groo
06-14-2001, 11:50 AM
I do about 110+ gigs of transferring a day with HTTP with no problems.

If you were to use FTP there is no way to know if somebody is stealing your links... with http it's trivial to set up a cron job to parse logs and check for outside linkers...

If you want to know why you cannot only allow downloads if the referrer is coming from your own domain -- the reason is this -- right-click/save target as will send a blank referrer -- and since these are large videos/music files then you I need to allow users to resume their files (with download managers, etc.). If you left-click a link to an mp3 then it will (more often than not) just open up winamp and start playing the file leaving the user with no means of saving it -- and novices don't know of right-click suprisingly. :)