[tahoe-dev] using the volunteer grid (dogfood tasting report)

zooko zooko at zooko.com
Mon Apr 6 23:17:56 UTC 2009


Folks:

I wanted to move some .flac music files from one of my laptops to my  
Linux workstation from which I could play them to my amplifier.   
Wait, I thought, instead of scp'ing them to the workstation, why not  
store them in a Tahoe grid where I can play them from anywhere?

So I did the following steps to join the volunteer grid:

1.  Find the furl for the volunteer grid introducer in the tahoe-dev  
mailing list archives.  (Hint, it is in this message, but the mailing  
list archives replace the at-sign with the string " at " and you have  
to undo that.  See also #649.)

2.  Get some version of the Tahoe source code (I did it with darcs to  
get the latest trunk) and run "python ./setup.py build".

3.  mkdir ~/.tahoe-testgrid && cd ~/.tahoe-volunteergrid

4.  $PATH_TO_SRC/bin/tahoe create-client .

5.  vi tahoe.cfg
  a.  nickname = draco (Zooko's Mac/PPC 867 MHz laptop)
  b.  web.port = tcp:9797:interface=127.0.0.1
  c.  introducer.furl = pb:// 
6cypm5tfsv6ag43g3j5z74qx5nfxa2qq at 207.7.131.41:64228,nooxie.zooko.com: 
64228/introducer
  d.  shares.needed = 4
  e.  shares.total = 6

I changed the web.port so that the node connected to the volunteer  
grid wouldn't try to get the same port number as other nodes I run on  
this machine.

I changed shares.total to be 6 because there are currently storage  
servers on the volunteergrid operated by six different people  
(including draco itself): SECORP (Peter), stockrt, draco (me), trid0,  
francois1, and ndurner.  Peter is contributing four different  
machines.  Unfortunately since server selection is random, any file  
could get 1 share on each of Peter's four machines.  Peter could then  
cackle cruelly and turn off those four machines and cause that file  
to become unavailable.  What I really want is to tell my node to  
upload one share to one machine per human operator, but that will  
have to wait until someone factors out the server selection logic so  
that I can script rules like that.

Then I started my node:

6.  $PATH_TO_SRC/bin/tahoe start .

7.  Then I pointed my web browser at 127.0.0.1:9797 and clicked  
"Create directory", then I started uploading my .flac files into that  
directory.

Yay!  Thanks for the storage service, Volunteer Grid contributors!  :-)

Regards,

Zooko

tickets mentioned in this email:
http://allmydata.org/trac/tahoe/ticket/649 # Validation of  
configuration settings



More information about the tahoe-dev mailing list