Description of problem:
Write/Append to a full volume causes fuse client to crash 

Steps to Reproduce:
1. Create a 1x2 volume 
[root@fedora2 /]# gluster vol info
 
Volume Name: test
Type: Replicate
Volume ID: 538b186b-b213-49f5-a5d9-458a3ba2690a
Status: Started
Number of Bricks: 1 x 2 = 2
Transport-type: tcp
Bricks:
Brick1: fedora2:/home/ssd/small_brick0/s0
Brick2: fedora2:/home/ssd/small_brick1/s1
Options Reconfigured:
performance.readdir-ahead: on
[root@fedora2 /]# 


with each brick to be 1 GB

2. on a fuse mount issue this command

for i in {1..1024}; do cat /var/log/journal/49001fda54c6e148ba137302d159e716/system >> file1; done

The input file is of 8MB and soon the volume will be full.

3. You get NO Disk space error and then the fuse client crashes.

(gdb) bt
#0  0x00007f627da49d4e in wb_do_unwinds (wb_inode=0x7f626c004170, lies=0x7f62776fc7a0) at write-behind.c:1069
#1  0x00007f627da4aae5 in wb_process_queue (wb_inode=0x7f626c004170) at write-behind.c:1453
#2  0x00007f627da493ad in wb_fulfill_cbk (frame=0x7f626800c0ec, cookie=0x7f626801150c, this=0x7f627800d970, 
    op_ret=-1, op_errno=28, prebuf=0x7f6268026964, postbuf=0x7f62680269d4, xdata=0x0) at write-behind.c:920
#3  0x00007f627dcacb83 in dht_writev_cbk (frame=0x7f626801150c, cookie=0x7f626801194c, this=0x7f627800c540, 
    op_ret=-1, op_errno=28, prebuf=0x7f6268026964, postbuf=0x7f62680269d4, xdata=0x0) at dht-inode-write.c:106
#4  0x00007f627def5b42 in afr_writev_unwind (frame=0x7f626801194c, this=0x7f627800af50) at afr-inode-write.c:204
#5  0x00007f627def5bd7 in afr_transaction_writev_unwind (frame=0x7f6268011bdc, this=0x7f627800af50)
    at afr-inode-write.c:221
#6  0x00007f627df044a6 in __afr_txn_write_done (frame=0x7f6268011bdc, this=0x7f627800af50) at afr-transaction.c:85
#7  0x00007f627df0aaa5 in afr_unlock_common_cbk (frame=0x7f6268011bdc, cookie=0x1, this=0x7f627800af50, op_ret=0, 
    op_errno=0, xdata=0x0) at afr-lk-common.c:633
#8  0x00007f627df0acb2 in afr_unlock_inodelk_cbk (frame=0x7f6268011bdc, cookie=0x1, this=0x7f627800af50, op_ret=0, 
    op_errno=0, xdata=0x0) at afr-lk-common.c:674
#9  0x00007f627e16cc0d in client3_3_finodelk_cbk (req=0x7f626000129c, iov=0x7f62600012dc, count=1, 
    myframe=0x7f626000119c) at client-rpc-fops.c:1676
#10 0x00007f628b22c1af in rpc_clnt_handle_reply (clnt=0x7f6278033c10, pollin=0x7f626801b690) at rpc-clnt.c:759
#11 0x00007f628b22c625 in rpc_clnt_notify (trans=0x7f62780340a0, mydata=0x7f6278033c40, 
    event=RPC_TRANSPORT_MSG_RECEIVED, data=0x7f626801b690) at rpc-clnt.c:900
#12 0x00007f628b228a1b in rpc_transport_notify (this=0x7f62780340a0, event=RPC_TRANSPORT_MSG_RECEIVED, 
    data=0x7f626801b690) at rpc-transport.c:541
#13 0x00007f628064cb3c in socket_event_poll_in (this=0x7f62780340a0) at socket.c:2231
#14 0x00007f628064d06a in socket_event_handler (fd=11, idx=1, data=0x7f62780340a0, poll_in=1, poll_out=0, poll_err=0)
    at socket.c:2344
#15 0x00007f628b4c0a24 in event_dispatch_epoll_handler (event_pool=0x1897d70, event=0x7f62776fcea0)
    at event-epoll.c:571
#16 0x00007f628b4c0dc6 in event_dispatch_epoll_worker (data=0x7f6278033800) at event-epoll.c:674
#17 0x00007f628a2b752a in start_thread () from /lib64/libpthread.so.0
#18 0x00007f6289c0622d in clone () from /lib64/libc.so.6
(gdb) 


Actual results:

Fuse client shouldnt crash

Expected results:

Fuse client crashes

Additional info:

Always reproducible.