Running Config Advisor against Clustered ONTAP Systems for Field Engineers

Running Config Advisor against Clustered ONTAP systems only presents the option of using the network - it is not possible to run Config Advisor over a Serial Port!

Image: Config Advisor 3.4 - Clustered Data ONTAP and Network tab

There are two options to manage this depending on the scenario:

1) Ask the customer to run config advisor!

Assuming Config Advisor is installed somewhere in the customer’s environment, and they have connectivity to a node-mgmt IP or cluster-mgmt IP, and cluster interconnect switches (if not single-node or switchless)

2) If the Field Engineer is equipped with a portable switch (like the below USB powered 8 port mini 10/100 metal switch from eten):

Image: Eten 10/100 USB powered 8 port mini switch

Then it can be done in the datacenter by connecting:

- laptop to switch
- switch to e0M port on a FAS controller
- switch to cluster interconnects admin ports (if not single-node or switchless)
- switch to management switches admin ports (only if using NetApp provided management switches)

Note: If this is a switchless cluster, or single-node cluster, then using a crossover cable to e0M would suffice.

A few considerations:

Laptop IP, a node-mgmt IP, and cluster/management switches’ admin IPs (if in use) will need to be on the same subnet.

Suggestion 1: Migrate node-mgmt LIF to e0M (if node-mgmt is on the same subnet as cluster interconnect switches):

Step 1: Plug in mini-switch to e0M

Step 2: Migrate the node-mgmt LIF to the e0M port:

::> net int show -lif mgmt1 -vserver NODENAME -instance
::> net int modify -lif mgmt1 -vserver NODENAME -failover-policy disabled -auto-revert false
::> net int migrate -lif mgmt1 -vserver NODENAME -destination-node NODENAME -destination-port e0M

Step 3: Run config advisor against NODENAME’s mgmt1!

Step 4: Put the node-mgmt LIF back where it was too:

::> net int revert -lif mgmt1 -vserver NODENAME
::> net int modify -lif mgmt1 -vserver NODENAME -failover-policy nextavail -auto-revert true

Note: Failover-policy = nextavail and auto-revert = true are the node-mgmt LIF defaults.

Suggestion 2: Create a temporary node-mgmt LIF on e0M (if existing node-mgmt LIF is on a different subnet to the cluster interconnect switches):

Step 1: Plug in mini-switch to e0M

Step 2: Create a temporary node-mgmt LIF on the e0M port on the same subnet as the cluster interconnect switches:

::> net int create -lif mgmt2 -vserver NODENAME -role node-mgmt -home-node NODENAME -home-port e0M -address X.X.X.X -netmask X.X.X.X -failover-policy disabled -auto-revert false

Step 3: Run config advisor against NODENAME’s mgmt2!

Step 4: Delete the temporary node-mgmt LIF

::> net int delete -lif mgmt2 -vserver NODENAME

Final Notes:

1) Config Advisor works against a cluster-mgmt IP
2) Config Advisor works against a node-mgmt IP (and a temporarily created second node-mgmt IP)
3) If you try and run Config Advisor against a switched cluster but choose the option “no switches” it will error with:

Image: Config Advisor Error “Execute … network options switchless-cluster … to enable switchless cluster”!

4) Config Advisor works against single-node clusters.
5) Cluster setup does not have to have been run to run Config Advisor (you might get some odd results though)!

To run Config Advisor against a node that is yet to be joined to a cluster:

Step 1: Abort the "Cluster Setup Wizard" by typing "exit" or using Ctrl+C, then run the following commands to create an IP Address for node-mgmt on the node-mgmt port e0M.

::> network interface create -lif mgmt1 -role node-mgmt -home-node localhost -home-port e0M -address X.X.X.X -netmask X.X.X.X
::> security login password -username admin
::> security login create -username admin -application ssh -authmethod password -role admin

Step 2: (Optional) Run the below commands if not connecting to the cluster interconnects at this time, or it is going to be a switchless cluster:

::> set advanced
::*> network options switchless-cluster modify -enabled true

Step 3: Run Config Advisor against the node’s mgmt1!

Step 4: (Optional) Unset the switchless-cluster option if this is not going to be switchless:

::> set advanced
::*> network options switchless-cluster modify -enabled false

Note: Single-node clusters actually have the switchless-cluster option set to false!

THE END!

Note: In this article I was testing against Clustered ONTAP 8.2.1RC1 SIMs - a single-node cluster, switchless cluster, and switched cluster!

Comments