[Gluster-users] bailout after period of inactivity

[mki-glusterfs at mozone.net]

On Tue, Jun 23, 2009 at 07:22:31AM -0500, Vikas Gorur wrote:
> > So there is a timeout, but whatever the cause is, it's triggered by
> > long term inactivity. We never had any network problems.
> > 
> > Other machines that access the filesystem on a regular basis do not
> > show this problem. It's only the machine that get's used once in a while.
> > The problem is reproducable, not a one time event.
>  
> Thanks everyone for the reports. We will try to reproduce this and resolve
> the issue.

Perhaps adding SO_KEEPALIVE may help maintain the socket connection in the
event that there's some form of tcp session timeout happening on the
network and/or router (such as in the case of NAT)?

Just a thought.

Mohan