There is a memory-leak in locks in glusterfs-3.0.5rc6(observed in 3.0.5rc5 too). 

valgrind report:
1,498 (1,344 direct, 154 indirect) bytes in 14 blocks are definitely lost in loss record 78 of 89
==4027==    at 0x4C2414B: calloc (in /usr/lib64/valgrind/vgpreload_memcheck-amd64-linux.so)
==4027==    by 0x663C35D: new_entrylk_lock (entrylk.c:43)
==4027==    by 0x663CCFA: __lock_name (entrylk.c:330)
==4027==    by 0x663DDF6: pl_common_entrylk (entrylk.c:660)
==4027==    by 0x663E07A: pl_entrylk (entrylk.c:720)
==4027==    by 0x4E4A16A: default_entrylk (defaults.c:1142)
==4027==    by 0x6A6DEF9: server_entrylk_resume (server-protocol.c:4538)
==4027==    by 0x6A75B44: server_resolve_done (server-resolve.c:594)
==4027==    by 0x6A75C35: server_resolve_all (server-resolve.c:629)
==4027==    by 0x6A75AE4: server_resolve (server-resolve.c:576)
==4027==    by 0x6A75C10: server_resolve_all (server-resolve.c:625)
==4027==    by 0x6A75935: server_resolve_inode (server-resolve.c:509)