Comment 36 for bug 932687

Ok, all my servers are VM hosted on AWS EC2.

I have one NFS server (called "back") that I recently upgraded on Ubuntu 13.04 with the saucy 3.9.2 kernel (I haven't tried 3.9.3 yet).
I have 1,n web server(s) (called "front") that runs PHP files that read/write media and cache files on the NFS server. "front" servers actually scales up and down, according to CPU usage.

I can reproduce the bug at will as every time autoscaling adds at least 1 "front" server, I get the bug. So I have to block autoscaling to have only one "front" server in order to work.

Actually I can't have only one server in production state, so I switched back on NFSv3.

I don't use kerberos. Here is some configuration files :

root@back:~# cat /etc/default/nfs-kernel-server
RPCNFSDCOUNT=8
RPCNFSDPRIORITY=0
RPCMOUNTDOPTS=--manage-gids
NEED_SVCGSSD=
RPCSVCGSSDOPTS=
RPCNFSDOPTS=

root@back:~# cat /etc/exports
/srv/exports *(rw,async,fsid=0,no_subtree_check,no_root_squash)
/srv/exports/data *(rw,async,no_subtree_check,no_root_squash)

root@front:~# cat /etc/fstab
LABEL=cloudimg-rootfs / ext4 defaults,noatime 0 0
/dev/xvdf /srv/apps ext4 defaults,noatime 0 0
## This is the NFS mount ##
10.0.0.254:/data /srv/nfs nfs4 defaults,rsize=32768,wsize=32768,noatime 0 0
###################
/dev/xvdb /mnt auto defaults,noatime,nobootwait,comment=cloudconfig 0 2