[volunteergrid2-l] I'm live. :)

[Shawn Willden]

On Wed, Apr 4, 2012 at 11:02 PM, Aaron Drew <aarond10 at gmail.com> wrote:

> Hey list,
>
> That was painless! From registered to live in about 2 minutes. Now
> uploading some data. :)
>

Awesome!

Up to 17 active nodes!

With 17 nodes in the grid (assuming Aaron's is stable, and I don't see any
reason why not to expect that), I would suggest the following configuration:

shares.total = 17
shares.happy = 15
shares.needed = 8

This provides a maximum expansion factor of just over two and a "happy"
expansion factor of just under two.  Assuming 95% node availability, this
provides a 99.9998% probability of file recoverability, on a per-file
basis, assuming only 15 shares are deployed.  Higher if all 17 are deployed.

I'd really like to use:

shares.total = 17
shares.happy = 17
shares.needed = 10

That would provide a consistent expansion factor of 1.7 with a 99.9999%
probability of file recoverability.  Unfortunately, there's some
hard-to-reproduce bug that causes shares.total=shares.happy configs to
occasionally fail to upload.  So until that's fixed it seems better to set
shares.happy = shares.total - 2 (shares.total - 1 also sometimes triggers
the problem, probably when some node is temporarily down).

When we get three more nodes, I'll suggest:

shares.total = 20
shares.happy = 18
shares.needed = 11

That will maintain a 99.9999% chance of file recoverability, while lowering
expansion to between 1.63 and 1.81.

Then our next goal is 30 nodes!

-- 
Shawn
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://tahoe-lafs.org/cgi-bin/mailman/private/volunteergrid2-l/attachments/20120405/e312f0b4/attachment.html>