[Gluster-users] NFS export issue with GlusterFS 3.1

[Shehjar Tikoo]

Bernard Li wrote:
> Hi Shehjar:
> 
> The hosts with exports in the replication pool are already running NFS
> servers so I need to setup GlusterFS native NFS server on another
> server.  I am using the following /etc/glusterfs/glusterfsd.vol:

Thats not recommended but I can see why this is needed. The simplest way 
to run the nfs server for the two replicas is to simply copy over the 
nfs volume file from the current nfs server. It will work right away. 
The volume file below will not.

Performance will also drop because now both your replicas are another 
network hop away. I guess the ideal situation would be to allow gnfs to 
run even when there is already a server running. Its on the ToDo list.

-Shehjar


> 
> volume gluster01
>     type protocol/client
>     option transport-type tcp
>     option remote-host gluster01
>     option remote-port 6996
>     option remote-subvolume brick
> end-volume
> 
> volume gluster02
>     type protocol/client
>     option transport-type tcp
>     option remote-host gluster02
>     option remote-port 6996
>     option remote-subvolume brick
> end-volume
> 
> volume repl
>     type cluster/replicate
>     subvolumes gluster01 gluster02
> end-volume
> 
> volume writebehind
>     type performance/write-behind
>     option cache-size 4MB
>     subvolumes repl
> end-volume
> 
> volume dshare
>     type performance/io-cache
>     option cache-size 1GB
>     subvolumes writebehind
> end-volume
> 
> volume nfs-server
>     type nfs/server
>     subvolumes dshare
> end-volume
> 
> and I start glusterfsd
> 
> Does this look about right?  Is the remote-port correct?
> 
> Thanks,
> 
> Bernard
> 
> On Tue, Oct 26, 2010 at 10:14 PM, Shehjar Tikoo <shehjart at gluster.com> wrote:
>> Bernard Li wrote:
>>> On Tue, Oct 26, 2010 at 9:15 PM, Shehjar Tikoo <shehjart at gluster.com>
>>> wrote:
>>>
>>>> Regarding this pdf, only the portions which show mount commands and the
>>>> FAQ
>>>> section is applicable to 3.1. In 3.1, NFS gets started by default for a
>>>> volume started with the volume start command.
>>> So basically you're saying if I have a 2 server replicated volume on
>>> gluster01 and gluster02, and I did volume start on gluster01, I should
>>> be able to mount the volume via NFS from either of the servers?
>>>
>> Yes. You can test that both servers have exported the same volume by:
>>
>> $ showmount -e gluster01
>> $ showmount -e gluster02
>>
>>> Is exporting via glusterfsd still supported?
>>>
>> Yes.
>>
>> -Shehjar
>>
>>> Cheers,
>>>
>>> Bernard
>>