[Gluster-users] Client hangs during server reboot (2-node replicated setup)

[Eric Mortensen]

Hello! 

I created a 2-node replica cluster with:

Volume Name: gsfiles
Type: Replicate
Volume ID: e01f6dc3-eb73-4bea-a187-eda98fe2748a
Status: Started
Number of Bricks: 1 x 2 = 2
Transport-type: tcp
Bricks:
Brick1: 10.2.3.100:/glusterdata/files
Brick2: 10.2.4.100:/glusterdata/files
Options Reconfigured:
auth.allow: 10.2.3.100,10.2.4.100

/etc/fstab:
/dev/mapper/vg.files-lvfiles  /glusterdata  ext4 noatime,errors=remount-ro,data=writeback
10.2.3.100:gsfiles    /files   glusterfs  defaults  0 0 

Both 10.2.3.100 and 10.2.4.100 are clients as well as servers.

When I reboot 10.2.3.100, and try to access /files from 10.2.4.100 the latter shell hangs indefinitely until 10.2.3.100 is up again. If I first stop the glusterfs and glusterfsd services, then I can access the data on the other node while the server reboots.

How do I avoid this? How do I ensure the system as a whole serves client requests even if one node goes down? 

Help greatly appreciated,
Eric Mortensen