Description of problem:

==6064== 220 bytes in 4 blocks are definitely lost in loss record 180 of 304
==6064==    at 0x4A04A28: calloc (vg_replace_malloc.c:467)
==6064==    by 0x4C5B188: __gf_calloc (mem-pool.c:150)
==6064==    by 0x4C1F6CE: _dict_set (dict.c:278)
==6064==    by 0x4C1F88C: dict_set (dict.c:321)
==6064==    by 0x4C2344D: dict_set_str (dict.c:2140)
==6064==    by 0x93ABC5A: client_setvolume (client-handshake.c:1546)
==6064==    by 0x93ACB5E: client_dump_version_cbk (client-handshake.c:1844)
==6064==    by 0x4EB19FB: rpc_clnt_handle_reply (rpc-clnt.c:797)
==6064==    by 0x4EB1D98: rpc_clnt_notify (rpc-clnt.c:916)
==6064==    by 0x4EADE7B: rpc_transport_notify (rpc-transport.c:498)
==6064==    by 0x854E26F: socket_event_poll_in (socket.c:1686)
==6064==    by 0x854E7F3: socket_event_handler (socket.c:1801)
Version-Release number of selected component (if applicable):
on git head

How reproducible:
executed valgrind once

Steps to Reproduce:
1. put valgrind over nfs process
2. execute untar of linux-kernel tarball and iozone
3. 
  
Actual results:


Expected results:


Additional info: