[p2p-hackers] Peers running out of storage

[Jim McCoy]

On Aug 8, 2008, at 12:27 PM, Tiago Vignatti wrote:
>
> When a DHT user try to register a resource that should be stored in a
> given node A, but that node runs out of storage space for this
> particular resource, then what happens? Who is responsible for this
> issue, the DHT itself or the DHT application?

In an ideal world the full peer would return a failure for the write  
attempt and then clients would treat that peer (for writes) as if it  
were a down/unavailable host.  You then just re-run the storage  
placement code to get the new "best" node for storing the datum.

The client should _always_ be responsible for taking care of this  
problem.  Trying to add a hidden layer of application logic that does  
this on behalf of the client is begging for a world of pain when the  
clients start assuming something worked (because the dht told them so)  
and there is a failure of any sort on the handoff code hidden behind  
the dht API.

jim