[Gluster-users] Unexpected behaviour during replication heal

[Whit Blauvelt]

On Tue, Jun 28, 2011 at 11:31:26PM +0100, Darren Austin wrote:

> There is no sync between the two servers in the situation I outlined, and
> the client cannot trigger a self-heal as you suggest because the client is
> effectively dead in the water until it's forcibly killed and re-mounted.

A way to "fix" this would be to set up a monit test of the client's access
to the file system that, just if it should become non-operational, would
restart the client.

Pretty easy to do. But if it's going to be necessary, it ought to be
documented. That is an ugly failure mode you've found. On the other hand,
any daemon that's critical should be watched by monit, or something similar,
especially if the daemon's not mature yet.

Whit