Description of problem:

GlusterFS 3.12.3

glusterfind sessions are inconsistent.

If you create a session with `glusterfind create`, it exists in `glusterfind list` only on the same machine, not on the other machines in the volume.

But if you `glusterfind delete` a session, it deletes any session with that name across _all_ machines!

That's inconsistent.

If you `glusterfind delete` a session that doesn't exist on any machine, it exits nicely with `Invalid session sessionname`.

But i fyou `glusterfind delete` a session that only exists on a remote machine, then you get the following ugly/nonsensical error message about SSH private keys not existing:

    [root@10-0-0-2:~]# glusterfind delete mysession myvol

    10.0.0.1 - delete failed; stdout (including remote stderr):
    Warning: Identity file /var/lib/glusterd/glusterfind/mysession/myvol/mysession_myvol_secret.pem not accessible: No such file or directory.
    Permission denied (publickey,password,keyboard-interactive).

    stderr:

    Command delete failed in 10.0.0.1:/data/glusterfs/myvol/brick1/brick
    10.0.0.3 - delete failed; stdout (including remote stderr):
    Warning: Identity file /var/lib/glusterd/glusterfind/mysession/myvol/mysession_myvol_secret.pem not accessible: No such file or directory.
    Permission denied (publickey,password,keyboard-interactive).

    stderr:


How reproducible: Always


Steps to Reproduce:
1. `glusterfind create mysession myvol` on one machine in the cluster
2. `glusterfind delete mysession myvol` on another machine in the cluster

Actual results:

Strange error message

Expected results:

Either both glusterfind create and delete should have sessions be local to only one machine, or bew global across all machines.