NetScaler administrators experiencing the below error when running "sync gslb" or "show gslb syncStatus"
show gslb syncStatus

Last known status:
Sync Time: Mar 16 2015 12:13:9
Retrieving local site info: ok
Retrieving all participating gslb sites info: ok
Site-A [Master]:
        Getting Config: ok
 
Site-B[Slave]:
        unable to create ctl_dir /nonexistent/.libnet-openssh-perl
        Aborting ...

Prior to build 10.5 55_8 passwords for rpcnodes (NSIP, loopback, GSLB siteip) should be same for autosync to work. The issue was resolved after matching the rpcnode passwords of all the above entities.

Problem Cause

Prior to build 55_8 passwords for rpcnodes (NSIP, loopback, GSLB siteip) should be same for autosync to work.
This has been fixed as part of TSK0497338 and the fix should be available in 10.5 55_8+ builds. However fix for TSK0497338 requires NSIP and loopback rpcnode password to be same.

From the output of the show rpcnode it seems that NSIP and GSLB site ip rpcnode passwords are different. Since the fix for TSK0497338 is not available in your current build make sure that passwords for NSIP, GSLB site ip and loopback rpcnodes are same.