Am 23.08.2012 20:27, schrieb Ben Johnson:
On 8/23/2012 1:42 PM, Knacktus wrote:
Hi all,
I'm a newbie to Apache, so I hope my question makes any sense :-) ...
I have a Python web application which runs on a RedHat 6.3 Server. I'm
not allowed to install any third party packages or modules, which are
not in the RedHat standard repository.
My app needs to serve and recieve large static files. The file downloads
need to be authorized. Google pointed me to mod_xsendfile, which was
excatly what I was looking for. Unfortunately, I can't install it on my
server.
A bit of a corollary, but I recommend that you switch hosts and purchase
a VPS on which you can do anything you like. That said, I realize that
this may not be an option for whatever reason (client's server, etc.).
Is there any other way to achieve a similar functionality with the
standard apache modules?
Thanks,
Jan
While I agree that mod_xsendfile is the ideal approach, I've had no
problem sending files up to 2GB in length to the user-agent via PHP
script (and Python should be capable of the same).
While sending files much larger than 2GB is possible, in theory, there
are many obstacles that make doing so a challenge.
So, if your files exceed 2GB in length, I see your concern. Otherwise,
why not use "straight-up" Apache? If Apache cannot meet your
authentication and/or authorization needs adequately, it seems that you
have no choice but to use Python for the task, in which case you would
also use Python to pass the binary files (from a location outside
Apache's document root) to the user-agent.
In other words, what, exactly, is the problem?
First I have to say, that I haven't done any real benchmarks yet. The
file size is below 1 GB, but the number of related, subsequent requests
can be around 100 ... 500 files with sizes from about 100 KB to 50 MB.
Some really large ones are about 200 MB, but those are rare. (The files
are CAD files.)
Right now, I only anticipate a problem because I read about the
advantages of the x_sendfile module (the files are not read by the
application code, but transfered directly by Apache). My main concern is
that the Python processes are blocked by the file uploads.
Authorisation is checked per file using some logic and another
web-service. So I would really like to keep that in my Python app, as
it's all developed, and I can't even tell, if it's possible at all doing
it with some Apache magic (newbie ...).
Probably I have to go the direction you pointed out (passing the file
via Python). Would it be possible / advisable to use two distinct
applications with two Apaches on different ports? One for the files, the
other for meta-data? I have full control over the client, which is a
richt-client. So I can route the requests for static files to one port,
the other requests to another port.
Cheers,
Jan
Thanks,
-Ben
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscr...@httpd.apache.org
For additional commands, e-mail: users-h...@httpd.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscr...@httpd.apache.org
For additional commands, e-mail: users-h...@httpd.apache.org
