How to troubleshoot Autolocation (Node Referrals) in ONTAP

Autolocation (or node referrals) is a CIFS feature that can allow for a more optimal connection to a Vserver that hosts CIFS data in Clustered Data ONTAP. Set to 'off' by default due to known configurations, where this feature is not recommended. See the end of this article for further details on those configurations.
Assuming you have a 3-node cluster, the cluster will have (3) nblades and (3) dblades, along with several data LIFs to access data.

To keep it simple, assume each nblade and dblade will be named: nblade# and dblade# (with # being 1, 2 or 3).

If Vservers have been created on each of the individual nodes in the cluster, it is much more likely that you will get better performance if you access the data directly from the node in the cluster that is hosting the data. 

Example: If you have a Vserver that has data residing on dblade1, it would be more likely that you will see better performance if you access the data through nblade1. This is the concept of autolocation or node referral.

At a high-level, it works as follows:

1. A LIF for a Vserver receives a CIFS request from a client for data
2. The Cluster identifies the root share the client is connecting to and determines which dblade controls the storage for the CIFS share root
   • If the dblade is local to the node the request came in on, then serve the data normally
   • If the dblade is NOT local AND there exists a data LIF which is local to the target dblade, then provide the client with a referral to that LIF
3. The client then connects to the LIF specified by the referral and continues with its CIFS operations.
   In order to provide the referral to the client, the cluster utilizes Microsoft DFS to provide the client with the 'Autolocation' information.