Questions about replacing bricks in Replicate/Distributed Replicate volumes

[oldunclez]

Hi
I am running glusterFS service among two servers .

root@server1:~# gluster --version

glusterfs 3.7.6 built on Dec 25 2015 20:50:46
Repository revision: git://git.gluster.com/glusterfs.git
Copyright (c) 2006-2011 Gluster Inc. <http://www.gluster.com>

I am reading https://docs.gluster.org/en/latest/Administrator-Guide/Managing-Volumes/#replace-brick，
and have the following questions :

1. Using the gluster volume fuse mount (In this example: /mnt/r2) set up metadata so that data will be synced to new brick

Does it mean running the command "root@server1:~# mount -t glusterfs server1:/r2 /mnt/r2" ?

2. (In this case it is from Server1:/home/gfs/r2_1 to Server1:/home/gfs/r2_5)

Should it be Server2:/home/gfs/r2_1 instead of Server1:/home/gfs/r2_1 ?
Because Server2:/home/gfs/r2_1 is the mirror of the faulty brick Server1:/home/gfs/r2_0

3. Is step "set up metadata"(i.e. setfattr /getfattr) necessary?
   I have made a test , do not take this step , and finally it work.

4. After excuting replace-brickcommand , there is no data in new brick Server1:/home/gfs/r2_5.
   But after excuting gluster volume heal r2 full , it does have data .
   So , Is gluster volume heal r2 full necessary? But docs.gluster.org does not mention it , or do I miss something?

Thank you for your reply .