Expanded DHT history analysis, and ran some more tests on DHT behaviour.

- Tweaked query parameters (quicker discard, longer requery period).

The analysis suggests that in general the DHT is running okay with some minor issues.
 - BitDHT peers respond to ~50% of FIND_NODE queries. This means we drop them, 
   and so  some ONLINE friends, become UNREACHABLE friends.

This appears to be caused by too many FIND_NODE queries reaching BitDHT, and the 
peer starts dropping them. This appears to come from specific peers, which 
are not RS nodes. Proposed solution is to Track and Filter Over-Querying Peers.
(Only one query per peer every 30 seconds).

 - Should be more lenient on the exact query matches.
Allow some dropped messages.

 - Queries appear to run longer than necessary.
A better query completion criteria needs to be established.

 


git-svn-id: http://svn.code.sf.net/p/retroshare/code/trunk@6614 b45a01b8-16f6-495d-af2f-9b41ad6348cc

libbitdht/src/bitdht/bdstore.cc

@@ -80,7 +80,7 @@ int bdStore::reloadFromStore()
 			{
 				addr.sin_port = htons(port);
 				bdPeer peer;
-				//bdZeroNodeId(&(peer.mPeerId.id));
+				bdZeroNodeId(&(peer.mPeerId.id));
 				peer.mPeerId.addr = addr;
 				peer.mLastSendTime = 0;
 				peer.mLastRecvTime = 0;