[tahoe-dev] [tahoe-lafs] #778: "shares of happiness" is the wrong measure; "servers of happiness" is better

[tahoe-lafs]

#778: "shares of happiness" is the wrong measure; "servers of happiness" is
better
--------------------------------+-------------------------------------------
 Reporter:  zooko               |           Owner:  kevan
     Type:  defect              |          Status:  new  
 Priority:  critical            |       Milestone:  1.5.1
Component:  code-peerselection  |         Version:  1.4.1
 Keywords:  reliability         |   Launchpad_bug:       
--------------------------------+-------------------------------------------

Comment(by zooko):

 Great!  Looking over your patch, this part catches my eye:

 {{{
 -            if placed_shares < self.shares_of_happiness:
 +            servers_with_shares = self._servers_with_shares()
 +            if len(servers_with_shares) < self.servers_of_happiness:
 }}}

 I guess this is the critical part of this patch.

 If the number of landlords falls below H, then we definitely won't get at
 least K-out-of-H reliability at the end, so it should indeed treat this as
 a failure here.  But, what if {{{len(servers_with_shares) >=
 self.servers_of_happiness}}}, and yet we still don't have K-out-of-H
 reliability because some of the servers have the same shares as each other
 (instead of different share numbers).  Can that happen?  I think so from
 reading the patch -- maybe we should have a test of this case.

-- 
Ticket URL: <http://allmydata.org/trac/tahoe/ticket/778#comment:51>
tahoe-lafs <http://allmydata.org>
secure decentralized file storage grid