[Gluster-devel] Being Gluster NFS off

[Jiffin Tony Thottan]

Hi all,

I am trying to list out  glusterd issues with the 3.8 feature "Gluster 
NFS being off by default".

As per current implementation,

1.) On a freshly installed setup with 3.8/3.9, if u create a volume, 
then Gluster NFS won't

      come by default and in the vol info we can see " nfs.disable on"

2.) For existing volumes(created in 3.7 or below), there are two 
possibilities

a.) if there are only volumes with default configuration, Gluster NFS 
won't come up and in

      vol info "nfs.disable on" won't displayed. In volume status 
command pid of Gluster NFS

     will be N/A.

b.) if there is a volume with explicit "nfs.disable off" set , then 
after upgrade Gluster NFS will

come and export all the existing volumes and vol info will have similar 
value as a.)

Currently three bugs[1,2,3] have opened to address these issues.

As per 3.8 release note, gluster nfs should be up for all existing 
volumes with default

configuration. We are planning to change this behavior from 3.9 onwards 
and Atin send out a patch[4]

With his patch after upgrade all the existing volumes with default 
configuration will have

nfs.disable value set to on explicitly in the vol info. So Gluster NFS 
won't export that volume at all

and gluster v status does not display status of Gluster NFS server.

This patch also solves bugs 2 and 3 as well

[1] https://bugzilla.redhat.com/show_bug.cgi?id=1383006 - gluster nfs 
not coming for existing volumes on 3.8

[2] https://bugzilla.redhat.com/show_bug.cgi?id=1383005 - getting n/a 
entry in volume status command

[3] https://bugzilla.redhat.com/show_bug.cgi?id=1379223 - nfs.disable: 
on" is not showing in Vol info by default
                               for the 3.7.x volumes after updating to 
3.9.0

[4] http://review.gluster.org/#/c/15568/

Regards,
Jiffin