RHGS 3.X : glusterd service refuses to start when volumes have options with very long lines

Issue

Starting the glusterd service fails while trying to update and restore volume information.
For example, on RHEL7/systemd, it shows up as the following :

# systemctl restart glusterd.service 
Job for glusterd.service failed. See 'systemctl status glusterd.service' and 'journalctl -xn' for details.

The glusterd logs, /var/log/glusterfs/etc-glusterfs-glusterd.vol.log, shows the following

I [glusterd-store.c:2075:glusterd_restore_op_version] 0-glusterd: retrieved op-version: 1
E [glusterd-store.c:2889:glusterd_store_retrieve_volume] 0-management: Failed to update volinfo for <VOLUME> volume
E [glusterd-store.c:3082:glusterd_store_retrieve_volumes] 0-: Unable to restore volume: <VOLUME>
E [xlator.c:406:xlator_init] 0-management: Initialization of volume 'management' failed, review your volfile again
E [graph.c:322:glusterfs_graph_init] 0-management: initializing translator failed
E [graph.c:538:glusterfs_graph_activate] 0-graph: init failed
W [glusterfsd.c:1183:cleanup_and_exit] (--> 0-: received signum (0), shutting down

[VOLUME] has an option set with a very long value (around 4096 characters or more). For example, auth.allow with a long list of IPs.

This happens either after assigning a volume option with a very long value :

# gluster volume <VOLUME> set <KEY> <VERY LONG VALUE>

Or upgrading from RHGS 2.1 to 3.X, since 2.1 did not have such a limitation.