Saturday, 18 May 2013

Preparing Nodes to Join a Pre-existing Clustered ONTAP Cluster

The following article covers steps to prepare nodes before joining an existing Clustered ONTAP Cluster. Here we go as far as preparing the nodes with hostname and management IP set, disks assigned, and SP (Service Processor) configured; prior to passing over for 'cluster join' and advanced cluster setup.
Note: Some of the steps below are optional extras, and the 'cluster join' is very easy.

1. Rack and Stack

Note: Be sure to check the shelf labels and keep stacks together when multiple stacks have been ordered.

2. Cable for Power

Note: Power supplies should all be turned off (‘0’).

3. Set shelf IDs

Power up the shelves and set the shelf IDs.
Generally, 00 is reserved for a FAS22XX internal shelf.
1X, 3X, 5X, 7X, 9X for the odd numbered node in a HA pair.
2X, 4X, 6X, 8X, 0X for the even numbered node in a HA pair.

4. Cable for SAS/FC disk shelf connectivity

Note: Please refer to the NetApp product documentation or the NetApp Commissioning Team’s excellent cabling diagrams.

5. Cable for ACP
6. Cable for HA
7. Cable to Cluster switches

Note: This can be left until just before the nodes are to be joined to the cluster.

8. Cable to Management switches

Note: This can be left until just before the nodes are to be joined to the cluster.

9. Cable Front End connectivity

For example: to CIFS, FC, FCoE, iSCSI, NFS … networks.
Note: This can be left until just before the nodes are to be place into production.

10. Controller power up

Monitor the output from both heads via a serial connection.
We let it boot into the Cluster Setup script, but, since we are not joining to the cluster here, we simply exit/quit the script. Default login is admin with no password.
Note: An option here is to first boot into Maintenance mode (Option 5 from the Ctrl-C Boot Menu) to first verify hardware status.

11. Set a hostname

system node rename -newname nodename

12. Set a node management IP

The below few lines ensure we don’t put node-mgmt on a port designated for cluster, and sets the management IP address and the default gateway. Here, by design, we’ll put node-mgmt on e0b and cluster-mgmt on e0a. Later, the node-mgmt will be configured to failover to e0a on the same node, and cluster-mgmt configured to failover to e0a on any nodes. It is recommended (but not necessary) to leave e0M - which is only a 100Mbps port - just for the Service Processor for various reasons such as backup and transfer speed benefitting by having node-mgmt on a 1Gbps port.

network port show
network port modify -node nodename -port e0a data
network port modify -node nodename -port e0b data
network port modify -node nodename -port e0c cluster -mtu 9000
network port modify -node nodename -port e0e cluster -mtu 9000

network interface create -lif mgmt1 -role node-mgmt -home-node nodename -home-port e0b -address X.X.X.X -netmask X.X.X.X

network routing-groups route create -routing-group nX.X.X.X/Y -destination 0.0.0.0/0 -gateway X.X.X.X

Note i: Y is the subnet mask (e.g. with 255.255.255.0, Y = 24). Clustered ONTAP tab completion will fill this for you.
Note ii: The routing-groups create command would usually have the -vserver switch; in this case - pre cluster join - there are no vservers.

To get SSH to the node management address working (if you don’t want to have to use the console):

security login password -username admin
security login create -username admin -application http -authmethod password -role admin
security login create -username admin -application ontapi -authmethod password -role admin
security login create -username admin -application ssh -authmethod password -role admin

13. Set the service processor (SP)

run -node nodename sp setup

14. Assign disks

set -priv adv
storage disk show -container-type unassigned
storage disk modify -disk nodename:XX.XX.* -owner nodename

From the other node can run:

::> system node run -node nodename
> disk assign all

15. Resize root volume (to avoid OCUM alert regards root aggregate full)

::> system node run -node nodename
> df -h
> df -A -h
> vol size vol0 250g
> exit

Note: See Hardware Universe for Minimum Root Volume sizes per FAS controller and DOT version.
Example: 250g is the minimum root volume size of a FAS6280 running CDOT 8.1.2.


16. Update/Verify BIOS Level


17. Update/Verify SP (Service Processor) Firmware Level


18. Update/Verify Disk Shelf Firmware


19. Update/Verify Disk Qualification Package


20. Update/Verify Disk Firmware


21. Update/Verify DOT Patch Level


22. Config Advisor (WireGauge) it


Ready for Cluster Join!

Further reading:
Check out this post from March 2013 - ‘How to Setup a Clustered ONTAP Cluster Without Using Cluster Setup Wizard’

No comments:

Post a Comment