[chord] Chord In PeerSim Problems

[Emil Sit]

[In the future, please continue threads on the mailing list.]

On Wed, 11 February 2009 at 12:44 (-0500), Karrels Daniel R CPT AFIT/ENG wrote:
> Thank you for your reply. I have also been using a known nodes list to
> assist in configuring the network. However, I am still uncertain how the
> protocol responds when a target node lies in an interval, however the
> preceding node is the host node. That is, given the node with ID 3172,
> described below, and a destination of 3196 (with no existing nodes in
> between), does the protocol dictate that the predecessor for the
> interval [3188,3204) is still node 3172? This is what I have so far, and
> this creates an infinite loop, even though the target node 3196 exists.

Fingers point at the first _succesor_ of id+2^i.

> If I set the fingers of a new node to point to its successor, then what
> I typically end up with is a livelock where the messages essentially
> bounce back and forth between the two nodes.

Lookups should terminate at the predecessor, who should return the
id of the successor to the querier.

Does that help?

-- 
Emil Sit / MIT CSAIL PDOS / http://pdos.csail.mit.edu/chord/