avoid loop between cluster gateways

The VIA extension is still needed to avoid some extra work and ugly
messages, but this is enough that it actually works.

This filters out the RemoteSides that are a proxied connection via a
remote gateway to the cluster.

The VIA extension will not filter those out, but will send VIA to them
on connect, which will cause the ones that are accessed via the listed
gateways to be filtered out.

Annex/Cluster.hs

@@ -96,8 +96,8 @@ clusterProxySelector clusteruuid protocolversion = do
   where
 	-- Nodes of the cluster have remote.name.annex-cluster-node
 	-- containing its name. Or they are proxied by a remote
-	-- that has remote.name.annex-cluster-node containing the cluster's
-	-- UUID.
+	-- that has remote.name.annex-cluster-gateway
+	-- containing the cluster's UUID.
 	isnode rs nodeuuids clusternames r = 
 		case remoteAnnexClusterNode (Remote.gitconfig r) of
 			Just names
@@ -106,11 +106,13 @@ clusterProxySelector clusteruuid protocolversion = do
 						ClusterNodeUUID $ Remote.uuid r
 				| otherwise -> False
 			Nothing -> case remoteAnnexProxiedBy (Remote.gitconfig r) of
-				Just proxyuuid -> not $ null $
-					filter (== clusteruuid) $
-					concatMap (remoteAnnexClusterGateway . Remote.gitconfig) $
-					filter (\p -> Remote.uuid p == proxyuuid) rs
-				Nothing -> False
+				Just proxyuuid
+					| Remote.uuid r /= fromClusterUUID clusteruuid -> 
+						not $ null $
+							filter (== clusteruuid) $
+							concatMap (remoteAnnexClusterGateway . Remote.gitconfig) $
+							filter (\p -> Remote.uuid p == proxyuuid) rs
+				_ -> False
 	
 	isclustername clusternames name = 
 		M.lookup name clusternames == Just clusteruuid

doc/todo/git-annex_proxies.mdwn

@@ -33,6 +33,9 @@ For June's work on [[design/passthrough_proxy]], remaining todos:
   protocol messages on to any remotes that have the same UUID as
   the cluster. Needs VIA extension to P2P protocol to avoid cycles.
 
+  Status: works, but needs VIA extension to avoid ugly messages and extra
+  work
+
 * Getting a key from a cluster currently always selects the lowest cost
   remote, and always the same remote if cost is the same. Should
   round-robin amoung remotes, and prefer to avoid using remotes that