[Gluster-devel] mount kernel-nfsd glusterfs-nfsd at same time

[Niels de Vos]

On Thu, Jun 13, 2013 at 09:59:40PM +0200, Heiko L. wrote:
> Hallo,
> 
> 
> we are run a backup-server on opensolaris with zfs exported per nfs and
> on this server should also run glusterfs-server exported per nfs.
> Both ways work successful separately, but not parallel.
> errormessage:
> 
>  l2 # mount -t nfs nfsbackup:/pool_s5_3/backup/l2/ /mnt/backup
>  mount.nfs: Unknown error 521
> 
> workaround:
>  nfsbackup # /etc/init.d/glusterd stop
> 
> It is possible to use both variants parallel?

No, not really conveniently and not without some limitations.

Only one NFS-server can be registered at the portmapper (MOUNT service 
provided by GlusterFS or rpc.mountd). On Linux you would use 'rpcinfo' 
to display which services the portmapper (portmap or the rpcbind daemon) 
knows. Depending on your NFS-client you can pass parameters to not 
contact portmapper to request the port for the MOUNT service, but 
contact the MOUNT service directly ('man 5 nfs' on Linux shows the 
'mountport' option).

Additionally the NLM protocol that handles the locking of files for 
NFSv3 registers itself to the portmapper too. Unfortunately I do not 
know if there are mount options for NFS-clients that can specify the 
port used for the NLM service. Mounting without locking could result 
into issues, but it could potentially be used as a workaround depending 
on your requirements and usage.

HTH,
Niels


> 
> Regards Heiko
> 
> details:
> 
>    (osol)   glusterfs-nfsd:38467 ------------ s1 (osol)
>   nfsbackup       osol-nfsd:2049 ------------ l2 (linux)
> 
> 
> nfsbackup # gluster volume status
> Status of volume: gv3
> Gluster process                                         Port    Online  Pid
> ------------------------------------------------------------------------------
> Brick node1:/zpool_tdedup/brick3/brick                  24011   Y       1034
> Brick node2:/pool_t3/brick3/brick                       24011   Y       2775
> NFS Server on localhost                                 38467   Y       1038
> Self-heal Daemon on localhost                           N/A     N       1040
> NFS Server on 192.168.5.172                             38467   Y       2777
> Self-heal Daemon on 192.168.5.172                       N/A     N       2779
> 
> 
> - glusterfs-nfsd-port: 38467
> - solaris-nfsd-port: 2049
> 
> 
> nfsbackup # gluster --version | head -1
> glusterfs 3.3.1 built on May 23 2013 21:45:52
> 
> nfsbackup # zfs get sharenfs  pool_s5_3/backup/l2
> NAME                        PROPERTY  VALUE                      SOURCE
> pool_s5_3/backup/elearning  sharenfs  rw=nfsclient,root=nfsclient  local
> 
> 
> s1 # mount nfs://nfsbackup:38467/$vol /mnt/$vol
> l2 # mount -t nfs nfsbackup:/pool_s5_3/backup/l2/ /mnt/backup
> 
> 
> 
> -----------------------------------------------------------
> [35] http://www2.fh-lausitz.de/launic/comp/sol/130430.osol.diag_make_glusterfs/130430.osol.make_glusterfs_howto.html
> 
> [39] http://www.gluster.org/pipermail/gluster-users/2010-December/029114.html
>    [Gluster-users] NFS mount with AIX, Solaris, HP-UX
>    2010
>    cmd : mount -o proto=tcp,vers=3 nfs://ylal3020:38467:/athena /users/glusterfs_mnt
> 
> 
> 
> _______________________________________________
> Gluster-devel mailing list
> Gluster-devel at nongnu.org
> https://lists.nongnu.org/mailman/listinfo/gluster-devel