[tahoe-dev] Benchmark and scaling figures and how to lockdown a tahoe-lafs grid...

[Shawn Willden]

On Sun, Nov 7, 2010 at 3:41 PM, Jimmy Tang <jcftang at gmail.com> wrote:

> this may sound like a silly question, I was reading through the
> configuration.txt file again to look up the M and K values, but which one is
> the M value?
>
> shares.needed = (int, optional) aka "k", default 3
> shares.total = (int, optional) aka "N", N >= k, default 10
> shares.happy = (int, optional) 1 <= happy <= N, default 7
>
> I get the impression M is the total number of shares (or machines?)
>

M is shares.total.  I thought the documentation used "M", not "N" for that
and "N" for the number of available storage servers.

The file is encoded into 'M' shares, such that any 'K' of them are
sufficient to reconstruct the file.  The M shares are distributed across the
available servers in a permuted order.  If there are less than M servers
available that will accept shares, the selection process "wraps" and goes
back to the first.  If there are less than H (shares.happy) servers
available to accept shares, the upload fails.  That last was recently
changed.  Before the change, Tahoe tried to ensure that it delivered at
least H shares, but it wouldn't matter if they were all on a single server.

-- 
Shawn
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://tahoe-lafs.org/pipermail/tahoe-dev/attachments/20101107/9b603589/attachment.html>