[tahoe-dev] servers-of-happiness default of 7 prevents first-time installation from working "out of the box"

[Zooko O'Whielacronx]

On Tue, Jun 15, 2010 at 6:32 PM, Brian Warner <warner at lothar.com> wrote:
>
>> At 2010-06-15 17:14 (-0600), Zooko O'Whielacronx wrote:
>>
>>> One possible solution to this would be to lower the default
>>> servers-of-happiness from 7 to 1. This would require us to also lower
>>> the default number of shares needed from 3 to 1, because the current
>>> code won't let you have a servers-of-happiness lower than your
>>> number-of-shares-needed:
>
> That's disappointing. I haven't looked closely at the
> shares-of-happiness changes, but I was hoping that it merely added a
> post-server-selection assertion that the shares were sufficiently
> distributed.
...
> I'd suggest removing the "SOH>=k" assertion and setting SOH=1 by default.

That was the idea when we created ticket #778 and Kevan began work,
but we quickly realized that there are subtleties to it so that we
couldn't give users sufficiently a unsurprising behavior that way

In particular, what would "servers of happiness" mean exactly, if we
did it that way? What it means now is "There are at least
$servers-of-happiness servers such than any subset of them of at least
size K can deliver your file.". If servers-of-happiness were less than
K then that definition wouldn't make sense, which would be fine if we
could come up with another definition which did make sense. Maybe "…
such that any subset of them of at least size K, or all of them if K
is greater than $servers-of-happiness, can deliver your file." ?

That ticket grew to be the champion ticket of all time so far on our
issue tracker -- two hundred and twenty eight comments from many
people over eleven months. Kevan wrote several implementations and
many iterations of docs and unit tests. One nice thing about those
docs and tests is that if you can come up with an alternate semantics
for what "servers of happiness" could mean then we should be able to
quickly tell whether the docs could be altered to explain those
alternate semantics and whether those alternate semantics would fail
one of the current tests.

Regards,

Zooko

http://tahoe-lafs.org/trac/tahoe-lafs/ticket/778# "shares of
happiness" is the wrong measure; "servers of happiness" is better