Saturday, 23 February 2013

Lab Series 02: Part 2 – (Clustered ONTAP) 2 Node Cluster Basic Elements

We’ve run through the setup and have our 2 Node Cluster (CLUSA) up and running, what elements make up this 2 node cluster?

The Elements

2 Cluster Nodes:
CLUSA-01, CLUSA-02

3 Vservers:
CLUSA, CLUSA-01, CLUSA-02

7 LIFs (Logical Interfaces):
- cluster_mgt on CLUSA (e0c on either node CLUSA-01 or CLUSA-02)
- clus1, clus2, mgmt1 on CLUSA-01 (ports e0a, e0b, e0f)
- clus1, clus2, mgmt1 on CLUSA-02 (ports e0a, e0b, e0f)

1 Failover Group:
clusterwide (containing 6 ports - e0c, e0d, e0e on CLUSA-01 & e0c, e0d, e0e on CLUSA-02)

2 Aggregates:
aggr0 on node CLUSA-01 (900MB, 96% used, raid_dp)
aggr0_CLUSA_02_0 on node CLUSA-02 (900MB, 96% used, raid_dp)

112 disks (after adding the optional SIM shelves):
- 4 shelves of 14 disks (56) assigned to each node
- 3 disks per node are contained in an aggr0 which holds the root volume
- 53 spare disks per node, of which there are
- 25 spare 1GB disks per node, and
- 28 spare 4GB disks per node

2 Volumes:
vol0 belonging to Vserver CLUSA-01 (851.5MB, 25% used)
vol0 belonging to Vserver CLUSA-02 (851.5MB, 25% used)

0 LUNs
0 igroups

A good grasp of the elements that make up a cluster, helps make the whole thing relatively easy to understand.

How to Find the Information?

i. Use OnCommand System Manager and click on Advanced which brings up the Data ONTAP Element Manager, or simply point your web browser at
http://MANAGEMENT_IP_ADDRESS_OF_CLUSTER or
http://MANAGEMENT_IP_ADDRESS_OF_A_CLUSTER_NODE

Image: Cluster Element Manager

ii. CLI

cluster show
vserver show
network interface show
network interface failover-groups show
aggr show
storage disk show
storage disk show -aggr aggr*
storage disk show -container-type spare
storage disk show -physical-size 3.93GB
volume show
df -h
lun show
lun igroup show

Three Re-Configurations

i. Removing e0d and e0e from the clusterwide failover group (used by the cluster management IP):

network interface failover-groups delete -failover-group clusterwide -node CLUSA-01 -port e0d,e0e
network interface failover-groups delete -failover-group clusterwide -node CLUSA-02 -port e0d,e0e
network interface migrate -vserver CLUSA -lif cluster_mgmt -dest-node CLUSA-02 -dest-port e0c

The above lines remove port e0d, and e0e from the failover-group clusterwide, so that the cluster management IP can exist only on e0c on CLUSA-01 or CLUSA-02. And will migrate the LIF cluster_mgmt to CLUSA-02’s port e0c.

Note: When the node homing the cluster_mgmt LIF is rebooted, the port fails over to a port on the other node.

ii. Renaming the aggregates containing the root volumes to have a matching naming convention:

storage aggregate rename aggr0 aggr_CLUSA1_00
storage aggregate rename aggr0_CLUSA_02_0 aggr_CLUSA2_00

iii. Modifying disk options to put autoassign off:

storage disk option show
storage disk option modify -node CLUSA* -autoassign off

Note: By default on the 8.1.2 C-Mode SIM, autoassign is turned on. In this lab all disks are assigned anyway, its useful practice though to check this is off.

No comments:

Post a Comment