943636fb   
   From: rick.jones2@hp.com   
      
   Madhur  wrote:   
      
   > I am observing this in one of the intermediate nodes. I had turned off   
   > lso for eth0. I am not sure if that is valid.   
      
   Is the intermediate node running a NIC which supports LRO - Large   
   Receive Offload (and has it enabled)?  There was a point in time in   
   the Linux space at least where cards had LRO, would aggregate   
   consecutive TCP segments they saw into larger segments, give them to   
   the host, and then the host, upon trying to forward them would find   
   them too large to send-out the next hop.  The workaround was to   
   disable LRO on the NIC (TSO or LSO is fine) of any intermediate system   
   - one that was an intermediate at layer 3 or below - application-layer   
   gateways were not negatively affected (IIRC) by LRO.   
      
   If this is indeed a JumboFrames issue, I didn't think that anything   
   out there (at least as an end-station) enabled JumboFrames by default.   
   The "rule" is that all stations in the same broadcast domain (ie   
   joined at layer 2 by bridges/switches rather than layer 3 with   
   routers) must have the same MTU.   
      
   rick jones   
   --   
   web2.0 n, the dot.com reunion tour...   
   these opinions are mine, all mine; HP might not want them anyway... :)   
   feel free to post, OR email to rick.jones2 in hp.com but NOT BOTH...   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)