BTW, the component is not NFS.

Ideally server-protocol should always allow 'NFS' server process to connect and work, and for NFS process client connection should be controlled by NFS layer, and not server-protocol layer.

When filtering the ip addresses with auth.allow, NFS requires the server ip addresses to be added specifically. This is a limitation, if we have hunderds of servers. Ideally 127.0.0.1 on one of the server should work.

Here is the scenario:

[root@centos5 export]# gluster volume set test auth.allow '192.168.1.77,127.0.0.1'
Set volume successful
[root@centos5 export]# showmount -e localhost
Export list for localhost:
[root@centos5 export]# gluster volume set test auth.allow '*'
Set volume successful
[root@centos5 export]# showmount -e localhost
Export list for localhost:
/test *
[root@centos5 export]# gluster volume set test auth.allow '192.168.1.77,127.0.0.1,192.168.1.95'
Set volume successful
[root@centos5 export]# showmount -e localhost
Export list for localhost:
/test *
[root@centos5 export]# 


[root@centos5 export]# gluster volume info

Volume Name: test
Type: Distribute
Status: Started
Number of Bricks: 2
Transport-type: tcp
Bricks:
Brick1: 192.168.1.95:/data/export
Brick2: 192.168.1.96:/data/export
Options Reconfigured:
diagnostics.brick-log-level: NONE
auth.allow: 192.168.1.77,127.0.0.1,192.168.1.95
[root@centos5 export]#

Planing to keep 3.4.x branch as "internal enhancements" release without any features. So moving these bugs to 3.4.0 target milestone.

Idea is to allow IP authentication in 'glusterfs' server-protocol to always allow for 127.0.0.1

Some hosts resolve to 127.0.1.1.

This option is supported and works fine:

Volume Name: new
Type: Distribute
Status: Started
Number of Bricks: 1
Transport-type: tcp
Bricks:
Brick1: sng:/export/dir2
Options Reconfigured:
auth.allow: 192.168.1.44,127.0.1.1

root@shishirng-laptop:~# showmount -e localhost
Export list for localhost:
/new *