[Gluster-users] Recovering from Arb/Quorum Write Locks

[Ravishankar N]

On 05/29/2017 10:45 AM, wk wrote:
> OK, can I assume SOME pause is expected when Gluster first sees 
> gluster2 go down which would unpause after a timeout period. I have 
> seen that behaviour as well. 

Yes, when you power off/shutdown/reboot a node, the mount hangs for a 
bit due to not receiving the socket disconnect notification immediately. 
Something to do with tcp keepalive timeouts which I am not familiar 
with. One way to avoid this while testing is to explicitly kill the 
bricks ( kill -15 <brick pid> ) on that node and then power it off. When 
you kill a brick, the mount receives the notification immediately.