Now that you have installed Heartbeat on Linux, you need to record key information about the cluster. In a Heartbeat cluster, most items that you add to the cluster are referred to as cluster resources.
|
HEARTBEAT CLUSTERING WORKSHEET |
|---|
|
Under Item 1: HA Management Client Password, record the password you want to use to provide access to the GUI HA Management Client for administering your Heartbeat cluster. Under Item 2: Nodes in the Cluster, list the nodes that you want to use for your GroupWise system. Identify each node by its DNS hostname and the IP address that it is physically configured with. Under Item 3: EVMS Container Resources, list the EVMS shared storage partitions that you want to use for GroupWise domains and post offices. An EVMS container resource needs a resource name and the name of an EMVS storage container. Under Item 4: File System Resources, for each EVMS shared storage partition, provide the necessary information about the file system on that partition. List the device name (for example,/dev/evms/gwsystem/gwvol1), mount directory (for example, /mnt/gwsystem), and file system type (for example, reiserfs or ext3) for the file system. In addition to the device, the directory, and the file system type, a file system resource also needs a resource name. |
NOTE:The reiserfs file system type is recommended for storing GroupWise data because it most efficiently handles the large numbers of small files that are stored in domains and post offices.
The number of nodes that are available in the cluster and the amount of shared storage in your SAN strongly influence where you can place GroupWise domains, post offices, and agents. You have several alternatives:
Your whole GroupWise system can run in a single cluster.
Parts of your GroupWise system can run in one cluster while other parts of it run in one or more other clusters.
Parts of your GroupWise system can run in a cluster while other parts run outside of the cluster, on non-clustered servers.
If you do not have the system resources to run all of your GroupWise system in a cluster, you must decide which parts have the most urgent need for the high availability provided by clustering. Here are some suggestions:
Post offices and their POAs must be available in order for users to access their GroupWise mailboxes. Therefore, post offices and their POAs are excellent candidates for the high availability provided by clustering.
Domains and their MTAs are less noticeable to users when they are unavailable (unless users in different post offices happen to be actively engaged in an e-mail discussion when the MTA goes down). On the other hand, domains and their MTAs are critical to GroupWise administrators, although administrators might be more tolerant of a down server than end users are. Critical domains in your system would be the primary domain and, if you have one, a hub or routing domain. These domains should be in the cluster, even if other domains are not.
The Internet Agent might or might not require high availability in your GroupWise system, depending on the importance of immediate messaging across the Internet and the use of POP or IMAP clients by GroupWise users.
The WebAccess Agent might or might not require high availability in your GroupWise system, depending on the importance of mailbox access from a Web browser.
There is no right or wrong way to implement GroupWise in a cluster. It all depends on the specific needs of your particular GroupWise system and its users.
In a GroupWise Heartbeat cluster, the required cluster resources for each resource group are an EVMS container, a file system, a secondary IP address, and one or more GroupWise agents. All of these types of resources, as members of a cluster resource group, fail over from one node to another while providing uninterrupted service to GroupWise users. So, for example, the resource group for a GroupWise domain would include the EVMS container and file system where the domain directory will be located and the secondary IP address that will be used by the MTA that services the domain. The domain directory and secondary IP address must be available before the MTA can start.
Typically, Heartbeat resources and groups are named with the object type first, followed by an underscore (_), followed by a unique name. For example, if you are creating a new GroupWise system and you want to put the primary domain in the same cluster resource group as the initial post office, you could name the resource group group_gwprimary. You could name the domain resource resource_gwmta and the post office resource resource_gwpoa. In this example, the domain and the post office would fail over together.
You might want the domain and post office together in the same cluster resource group or in different cluster resource groups. If you want the domain and the post office to fail over separately, you need two separate cluster resource groups. For the domain, you could create a resource group named group_domain1 and a resource named resource_mta1. For the post office, you could create a resource group named group_post1 and a resource named resource_poa1.
|
HEARTBEAT CLUSTERING WORKSHEET |
|---|
|
Under Item 6: Cluster Resource Groups for GroupWise Components, list the cluster resource groups that you want to create for GroupWise components. Initially, you plan one or two resource groups for the primary domain and initial post office. Later, you plan additional resource groups for additional GroupWise components. A GroupWise resource needs a resource name and the name of the domain or post office equivalent to its eDirectory™ object name. |
Each cluster resource group needs a unique secondary IP address in the cluster. The secondary IP address remains constant regardless of which node the cluster resource group is running on. It is independent of all physical IP addresses in your network.
|
HEARTBEAT CLUSTERING WORKSHEET |
|---|
|
Under Item 5: IP Address Resources, list the secondary IP addresses that you want to use for GroupWise agents. The IP address resource also needs a unique name. Initially, you plan one or two secondary IP addresses for the MTA and the POA for the primary domain and initial post office. Later, you plan additional secondary IP addresses for additional resource groups. |