Bug 965869

Description of problem:

Server is configured with replica 2 with 2 servers (say p01 and p02).

When one of the server is taken offline for maintenance, the client is supposed to be able to access the files. However, there was a symlink in particular that when a client tries to access, the client gets "Transport End point disconnected". Then, a re-try of ls on that symlink works fine shortly after.

Since only 1 of the servers is taken offline, ls -l should continuously work.

# gluster volume info puppet_data
 
Volume Name: puppet_data
Type: Replicate
Volume ID: 5796ee4a-b58f-49a7-8b45-b9f05174040b
Status: Started
Number of Bricks: 1 x 2 = 2
Transport-type: tcp
Bricks:
Brick1: p01:/mnt/gluster/puppet_data
Brick2: p02:/mnt/gluster/puppet_data
Options Reconfigured:
auth.allow: 10.0.86.12,10.0.72.135,127.0.0.1,10.0.72.132,10.0.72.133
nfs.disable: off
nfs.register-with-portmap: 1



Version-Release number of selected component (if applicable):
Server on Redhat 6.2 64-bit (could not upgrade to 3.3.1 because of NFS bug on 3.3.1 - filed with bugzilla):
glusterfs-fuse-3.3.0-1.el6.x86_64
glusterfs-3.3.0-1.el6.x86_64
glusterfs-server-3.3.0-1.el6.x86_64


Clients on Redhat 6.2 64-bit (2.6.32-220.el6.x86_64)
glusterfs-fuse-3.3.0-1.el6.x86_64
glusterfs-3.3.0-1.el6.x86_64


How reproducible:
Not very

Steps to Reproduce:
1. Create 2 Gluster Servers with replica 2 
2. Reboot one of the servers
3. ls -l a file mounted by the Gluster client

Actual results:
ls -l on any files hosted by Gluster should continuosly work; not producing "Transport Endpoint not connected error" (even though it recoveres)

Expected results:
ls -l on the file on Gluster volume should not return error

Additional info:

Happened to me couple other times.