SmartConnect and CName issue

what do we do to get it to use all nodes?

CNAME should point to NS record

> If we ping the cluster name it will cycle through the different IPs

Good, SmartConnect has been set up correctly.

> if we ping the CName it returns the same IP every time.

With CName you mean the address of the SmartConnect DNS service?

This is in fact a single fixed IP address which resides on the first node,

and it will fail over to the next node in case of a node outage.

hth

-- Peter

The CNAME is pointed to the NS Record.

> This is in fact a single fixed IP address which resides on the first node,

Technically, it's the node with the lowest device ID.  This may or may not be the node with the lowest logical node number.  I've got one cluster where it ended up being $clustername-18.  If you built your cluster from node 1 in order and never replaced a node, it will be the "first" node.

> what do we do to get it to use all nodes?

That's not how the Force works...

The SmartConnect name service is not designed as distributed

in the sense that it would run on all nodes at the same time.

It's considered sufficiently high-available with the aforementioned

fail-over mechanism from one single node to another one.

Try it out by shutting down your first node

(as Ed pointed out, the node with the lowest device ID, not logical number).

Check wether during the fail-over there is a time gap where the cluster name can't be

resolved in your environment (better use host or dig rather than ping).

-- Peter

Hi wilmhealth,

Can I ask why you need to use a CNAME? It isn't required for setting up SmartConnect. There is more information in the Best Practice Guide:

Page 19 discusses DNS delegation https://support.emc.com/docu58740

The only mention of CNAMS is in dealing with migration and it is not the preferred method (page 15).

We have 3 nodes and almost all traffic is going through 1 node and we would like to spread it across all nodes.

Use  your cluster name for mounting, not the CNAME...