Ideally, as rebalance is part of glusterfs internal activity, the 'gfid' (which is the identity for a file, like inode number) gets changed. This is not causing any problems as of now because glusterfs being distributed system, can do a re-lookup and get the info of file even if the gfid changes, but in theory, the gfid of the file should remain same even after rebalance. (which is very important to handle open-fd rebalance).
PATCH: http://patches.gluster.com/patch/7711 in master (distribute: bring in feature to do rebalance from inside)
PATCH: http://patches.gluster.com/patch/7712 in master (glusterd-rebalance: utilize distribute's internal rebalance feature)
I created a distribute volume with one brick and started it and mounted it as fuse mount. I created some 5 files using touch on the mount point and checked their gfids in the backend. Then added one brick to the volume and started rebalance,then checked gfids of all the files in the backend. Now gfids won't change with the rebalance.