> Sending a SYN will only happen during the time that client had a valid > ARP entry in its cache from a previous connection. As soon as the ARP > cache is invalidated/updated and there is no value address to resolve > the 'bad' mount would be time out faster (that's when all the other > mounts would be unblocked but the hanging mount would take longer to > retry a number of times). In addition to what Michael Wakabayashi already said, I would like to point out that if the NFS client is not in the same subnet as the NFS server, then the client would never have an ARP entry for the NFS server. Instead, the client would have the ARP entry for the default gateway of its subnet where it would happily send a SYN. This is exactly why we are asking you to re-create the issue with the exact command that Michael Wakabayashi has provided, as we believe that fake IP of 2.2.2.2 would be outside of your NFS client's subnet range. Thanks. -Alex