geo-replication becomes faulty because of root filesystem getting full

Issue

Geo-replication session between a master brick and a slave node can become faulty if the root filesystem on that node becomes 100% full. This is because the working directory of the geo-replication in a node is present in the root filesystem ( /var/lib/misc/glusterfsd//ssh.*/// to be precise). If the filesystem that hosts the working directory of geo-replication becomes full, then it can lead to a faulty geo-replication session.