How to add a volume into HA cluster

http://high-availability.com/FAQ/index.php?action=artikel&cat=13&id=40&artlang=en&highlight=Adding+a+volume

To add a volume to the cluster first select the Add a volume tab from the volume operations section. This presents a configuration similar to the one shown below. Note that volumes presented in the an style="font-family: arial,helvetica,sans-serif;">Volumean> drop down list are made up of volumes imported on the cluster node the GUI is being run on and are not already part of the cluster.

Once a volume is selected, the following information should be entered/set:

• Failover hostname

  The hostname to be associated with this volume service. This is the hostname that clients should use to connect to the service.

• Primary appliance

  Specifies the preferred node in the cluster to bring the service up on.

• Heartbeat devices

  This is a list of the virtual devices (vdevs) in the pool. As of NexentaStor 3.1 the cluster software is able to heartbeat directly throught these devices. By default the first two devices are selected - this is the recommended number of heartbeats per pool. You can select more heartbeats if required, change the two that are being used, have a minimum of one heartbeat or indeed specify all devices using the select all checkbox in the list heading. Note that each pool will have it's own set of heartbeat devices, so if the cluster has three pools each with two heartbeat devices each then a total of six heartbeat devices will be defined in the cluster.

• Failover interface

  For each node in the cluster a drop down box provides a list of the interfaces available to which the failover hostname is bound when the volume is running on that node.

There are a number of more advanced settings that can be altered by selecting the Advanced Settings tab. Note that these settings are not mandatory as sensible defaults have been selected for you.

alt="" width="907" height="517" />

• Netmask

  An optional netmask to be applied to the failover hostname. If one is not provided then the default netmask for the failover hostname network class is used.

• Initial timeout

  This is the initial time that cluster nodes wait to contact other cluster nodes when they first boot up. If this timeout expires the node will then start any services that are in automatic mode. However, if during this timeout other cluster nodes are contacted then the timeout is canceled and the nodes negotiate between themselves which services to start.

• Standard timeout

  The time after which it is determined contact is lost with another node. Once expired any services eligable to run on this node that are not running elsewhere in the cluster will be started.

• Enable SCSI-2 reservations

  Issue scsi reservations on the devices in a volume before importing it; this is done to enforce data integrity (i.e. prevents the pool being imported on two nodes at any one time). Under normal circumstances this option should be left enabled and should only be turned off if instructed to do so by Nexenta support staff.

• Volume export failure

  Should a volume export fail then this option forces the node the export is failing on to halt immediatly. This option was a requirment for legacy iSCSI and is no longer needed for COMSTAR.

Finally click the alt="" width="155" height="24" /> buttonto complete the volume addition.

If the failover hostname is undefined on any or all of the nodes in the cluster then a popup window will prompt you to enter the IP address of the hostname. The input field is aware of the correct format of IP version 4 addresses and will place a red line under an incomplete address thus:

alt="" width="342" height="217" />

Once a syntactically correctly address is added the red underline is removed and the hostname/address can be added to the cluster:

alt="" width="340" height="215" />

Finally a popup appears to confirm the volume addition:

alt="" width="401" height="127" />

Removing a volume

Click on the Remove a volume tab to be presented with a table of volumes in the cluster similar to the example below:

alt="" width="909" height="465" />

To remove a volume simply click the alt="" width="16" height="16" /> icon on the row of the volume you want to remove. A popup then confirms volume removal:

alt="" width="556" height="129" />

When removing a volume the following points should be noted:

• The volume will be removed from the cluster without effecting any other volumes in the cluster.
• The volume will be left imported on the node it is running on. If it is not running on any nodes in the cluster then it will be left exported.
• The failover hostname associated with the volume will be moved to NMS control.