SIGHUP doesn't do anything for 'glusterfs' client processes
anymore.

In GlusterFS 3.2.7, sending a SIGHUP signal to 'glusterfs'
client process would cause it to reincarnate(), which caused
the client process reread the volume spec file. This was
obvious by observing its log file.

Using SIGHUP was useful when the GlusterFS distribute volume
was expanded in size, as without the SIGHUP, the client
would not see the new size.

In GlusterFS 3.4.2, the SIGHUP to the 'glusterfs' client
process appears to do nothing at all, although the call to
reincarnate() is still there, and the signal does not appear
to be blocked.

So in GlusterFS 3.4.2, then the volume is changed in shape
or size, the 'glusterfs' FUSE mount must be unmounted, which
requires the applications using the mount to be stopped as
well -- kind of disruptive.

I hoped that the unmount might be avoided with the "mount -o
remount" option, but apparently FUSE mounts to not support
the 'remount' option at all.

It would be useful if the old 3.2.7 SIGHUP behaviour could
be returned, or better, if the client 'glusterfs' process
would learn of the new volume shape/size through
communications with the 'glusterfsd' process without
requiring any user manual actions at all.