[OpenAFS] afs directory

Rodney M Dyer rmdyer@uncc.edu
Sun, 10 Jul 2005 22:01:38 -0400

At 05:25 PM 7/10/2005, you wrote:
>To quote Guns and Roses, "I believe what we have here is a failure to 
>communicate".  Okay,

Ok first, that isn't a Guns and Roses quote, that line is from one of the 
sweatiest movies ever made "Cool Hand Luke".  Rock on.

>1.  I don't do windows

Good enough.  No problem.

>2.  I do have AFS clients and servers working both from RPM's and compiled.

Great, you're ahead of the game then.

>3.  I do have kerberos authentication enabled, set-up and working by my 
>lonesome self.

Alright, even better.

>4.  Other rants about being a self-made man that you don't have time for.....

We all have our axe's to grind.  :)

>5.  I have no issues with AFS itself, things are just fine.

Working client...  check!

>6.  Rather than spend hours searching AFS archives WITHIN AFS, is there a 
>more timely
>     sequence of events I can perform to find files within the GREATER AFS 
> filespace, i.e.

No, there isn't.  What you are asking for is more like Apple or Microsoft's 
new search engines, Spotlight, and WinFS respectively, that keep databases 
of what is on your local hard disk so you can perform a quick 
search.  There is no way to search AFS cells like searching a local hard 
drive, other than to do a bute force search for every file and directory.

>7.  If I have to ftp or google to find a file on some .edu server what the 
>heck do I need AFS for???

AFS is a network file system.  You use it to store files on the network, as 
opposed to your local disk.  Once stored there, you can retrieve the file 
later by remounting the same AFS cell and traversing its subdirectories to 
the point where the file is at.  You need to remember that path, just like 
you remember how to type in a web address.  What makes AFS different than 
the web is that AFS can be used directly by your OS applications when they 
need to open a file by using any of the traditional "file open" OS API 
calls.  You don't have to go "though" a web server to get to a file.  You 
can use "VI" to edit a file right out of AFS.  You can't do that with the 
web.  There are probably several thousands of terrabytes of on-line data 
stored in AFS.

People who run AFS cells do so because they need an enterprise wide common 
location for infrastructure data.  AFS isn't an "archive" as much as it is 
an active on-line "data repository".  The word "archive" to me seems to 
imply "backup".  AFS "can" be used to backup user data, but that isn't its 
primary purpose.  Understanding the difference between what AFS is meant 
for and what the Web is meant for is critical.  They are not the same beast 
and definitely serve different functions.  Also remember that the web 
typically serves "public" data verses AFS which is used to access an 
organizations internal private data as well as public.

Some people run web servers "on top" of AFS.  That means when you fetch a 
web page with your web browser, the file is actually being pulled out of 
AFS by the web server and sent to you.  Our site does that.

I believe what you are looking for might be more akin to the old "archie" 
UNIX utility for searching for files on a host.  Don't be so quick to 
dismiss the need for AFS.  You are asking to do something that AFS doesn't 
have a functional design for.  Apples are not oranges.