[Spread-users] Membership algorithm--continued

kevin Tian kevin_tian9844 at yahoo.com
Fri Apr 16 03:24:27 EDT 2004

> This corresponds to the case where a ring
> representative called 
> Lookup_new_members(), found that there were members
> not in its ring, 
> shifted to GATHER to try to find them, and then
> timed-out without 
> finding any other representatives to try to merge
> with.  Thus, this case 
> is simply a failed attempt to increase the size of
> the ring from a 
> currently healthy membership.
Got it.

> Ideally, the smallest rep will timeout first and
> send a form1 token 
> (because everyone else pushes their timeout back
> when they notice a new 
> smallest rep).  If this isn't the case, then the
> other reps can only 
> conclude that gather failed, and try again.  Of
> course, if there's just 
> a time delay here, this doesn't necessarily hurt the
> membership... 
> they'll handle the form1 if it comes.
Got it, as is done in Memb_handle_form1, just
fill_form1 in various state. 

> I'm not sure exactly what you mean, here. 
> Distributed agreement is 
> impossible in an asynchronous environment.  Could
Yes, it is. But it goes when network is in stable
runs, when failure detectors behave like eventually
perfect ones--- many papers explain it this way.:-)
> you explain what you 
> mean by violating the algorithm description? 
Here I refer to the membership algorithm description
in Dr. Amir’s PhD thesis.
He said, ‘In the Commit state the representatives
reach an agreement, they agree on the set of
representatives that will participate in the formation
of the new ring’
I do not think the codes detect this agreement after
GATHER timeout, instead, because the two-round
rotation of the form token is necessary for EVS and
ordering semantics, it can be done in the two-round
rotation of the form token---first fill members info,
then read them---this way, consensus is reached. 
BTW, Does F_members refer to the list of active daemon
in one segment?
I’ve just browsed a little of related codes, maybe I
am incorrect. 
If more notes, especially for some global variables
and data structures, are available, it will be fine:-)


Do you Yahoo!?
Yahoo! Tax Center - File online by April 15th

More information about the Spread-users mailing list