[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:  zooko   
     Type:  defect                     |          Status:  assigned
 Priority:  critical                   |       Milestone:  1.6.0   
Component:  code-peerselection         |         Version:  1.4.1   
 Keywords:  reliability review-needed  |   Launchpad_bug:          
---------------------------------------+------------------------------------

Comment(by zooko):

 replying to myself:

 Replying to [comment:156 zooko]:
 >
 > The algorithm for calculating servers-of-happiness starts with a map
 from servers to shares that the server holds. More than one server can map
 to a single share and more than one share can be mapped to from a single
 server.
 >
 > The algorithm looks for a share that has more than one server mapping to
 it and then excludes all but one of those links (from server to share).
 Then it iterates this until there are no more cases of a share with more
 than one server mapping to it, and then it is done and the happiness value
 is the number of servers mapping to a share.
 >
 > Now to finish fully defining the algorithm we have to explain how it
 chooses which link to retain when it finds multiple links pointing to a
 share.

 Oh, ''and'' we have to specify how it chooses which share to examine next.
 Presumably a good way to choose the next share plus a good what to choose
 which links to exclude can solve the puzzle I posted in comment:156 and
 other similar puzzles.

-- 
Ticket URL: <http://allmydata.org/trac/tahoe/ticket/778#comment:157>
tahoe-lafs <http://allmydata.org>
secure decentralized file storage grid