GeoCluster resource properties

There are six properties tabs for the GeoCluster Replicated Disk resource on Windows 2003.

1. General—This tab identifies the Name and Description of the resource and the Possible Owners. If you change the name of the resource, the replication set name will not change until the next time the resource is brought online.

Modifying the Possible Owners list will cause one of the following actions to occur.

• If you add additional Possible Owners, the GeoCluster Replicated Disk resource will connect the resource’s replication set to the new owners and begin a mirror to each.
• If you remove Possible Owners, the GeoCluster Replicated Disk resource will disconnect the resource’s replication set from each owner removed.

The GeoCluster Replicated Disk resource must have at least two possible owners to function properly.

2. Dependencies—By default, the GeoCluster Replicated Disk resource is not dependent on any other resources.
3. Advanced settings—This tab controls how and when MSCS handles a failure of the resource.

• Do not restart—Select this option if you do not want cluster service to restart the resource if it fails.
• Restart—Select this option if you want cluster service to restart the resource if it fails.
• Enable Affect the group if you want a failure of this resource to move the group to another node. If you disable this option, cluster service still attempts to restart the resource using the Threshold and Period values, but the failure of the resource will not cause the group to move to another node.
• The Threshold and Period values determine the number of times cluster service will attempt to restart the failed resource within a specified period of time before moving the group to another node.
• "Looks Alive" poll interval—This setting specifies how often the resource is polled to determine whether it is still running on the active node. You can choose a value from the resource type, or you can specify your own value.
• "Is Alive" poll interval—This setting designates how often the possible owners are polled to determine whether the specified disk on each node can be written to and read from. You can choose a value from the resource type, or you can specify your own value.
• Pending timeout—This value determines how long the resource is allowed to remain in a pending state before it fails. If the resource takes longer than the time specified to be brought online or taken offline, the resource fails.

For more information on the Advanced Settings options, see your Windows documentation.

4. Connection parameters—This tab controls disk replication, network routing, and orphan files for GeoCluster.

• Disk to replicate—The volume to replicate
• Network to route Double-Take mirroring and replication traffic over—The network to use for Double-Take Availability mirroring and replication traffic. If you do not have multiple networks established, you will only be able to select the one network that does exist. If you do not select a network, GeoCluster will use DNS to determine a network route to use.

Ideally, the networks used for various traffic should be separated. This is dependent on the number of networks that you established when you created the cluster and the priority assigned to each network. For example, if you have two network routes, separate GeoCluster and your public traffic. If you have three routes, separate the public traffic and then separate GeoCluster from the cluster heartbeat.

Modifications to either of the first two settings will not take effect until the next time the resource is brought online.

• Interval to check unresponsive nodes—The frequency to determine how often an unresponsive node is checked to see if a Double-Take Availability connection can be made
• Delay connection until resources dependent on this one are online—This option allows you to delay a Double-Take Availability connection until any resources that have the GeoCluster Replicated Disk resource as a dependency are online. By ensuring that all resources that are dependent on the GeoCluster Replicated disk resource are online before starting the connection, the chance of a conflict occurring because application resources are attempting to open files exclusively while GeoCluster is mirroring those files is removed.
• Orphan files—An orphan is a file that exists in the target location but is not in the source location. You can enable the resource to remove or delete orphans during a mirror. If you choose to move orphan files, specify the location where you want to move them. You should not select the system volume for the orphan files because it could impact the stability of the cluster service.

5. Compression—If you want to configure Double-Take Availability compression, verify that Enable Compression is selected. Depending on the compression algorithms available for your operating system, you may see a slider bar indicating different compression levels. Set the level from minimum to maximum compression to suit your needs.

6. Mirror Properties—This tab controls the Double-Take Availability mirroring process.

• Full Mirror—All files in the replication set will be sent from the source to the target.
• File differences—Only those files that are different based size or date and time will be sent from the source to the target.
• Send data only if Source is newer than Target—Only those files that are newer on the source are sent to the target.

Note:

If you are using a database application, do not use the newer option unless you know for certain you need it. With database applications, it is critical that all files, not just some of them that might be newer, get mirrored.

• Use block checksum—For those files flagged as different, the mirror performs a checksum comparison and only sends those blocks that are different.
• Calculate Replication Set size prior to mirror—Determines the size of the replication set prior to starting the mirror. The mirroring status will update the percentage complete if the replication set size is calculated.

There are eight properties tabs for the GeoCluster Replicated Disk resource on Windows 2008.

1. General—This tab identifies the Name and Resource type of the resource. It also displays the current state of the resource and an additional detailed status message.
2. Dependencies—By default, the GeoCluster Replicated Disk resource is not dependent on any other resources.
3. Policies—This tab controls how and when MSCS handles a failure of the resource.

• If resource fails, do no restart—Select this option if you do not want cluster service to restart the resource if it fails.
• If resource fails, attempt restart on current node—Select this option if you want cluster service to restart the resource if it fails. Specify the length of time to attempt restarts and the number of restarts to attempt during that period of time.
• If restart is unsuccessful, fail over all resources in this service or application—If this option is enabled, the failure of the group will cause the resource to move to another node. If this option is disabled, the failure of the resource will not cause the resource to move to another node.
• If all the restart attempts fail, begin restarting again after the specified period—If this option is enabled, the cluster will delay the length of time specified before trying to restart the resource again.
• Pending timeout—This value determines how long the resource is allowed to remain in a pending state before it fails. If the resource takes longer than the time specified to be brought online or taken offline, the resource fails.

For more information on Policies options, see your Windows documentation.

4. Advanced Policies—This tab controls resource specific settings.

• Possible owners—All nodes of the cluster are listed. Select or deselect the nodes that you want to be possible owners.
• If you add additional owners, the GeoCluster Replicated Disk resource will connect the resource’s replication set to the new owners and begin a mirror to each.
• If you remove owners, the GeoCluster Replicated Disk resource will disconnect the resource’s replication set from each owner removed.

The GeoCluster Replicated Disk resource must have at least two possible owners to function properly.

• Basic resource health check interval—This setting is formerly known as the Looks Alive poll interval. It specifies how often the resource is polled to determine whether it is still running on the active node. You can choose the standard time period of 5 seconds, or you can specify your own value.
• Thorough resource health check interval—This setting is formerly known as the Is Alive poll interval. It designates how often the possible owners are polled to determine whether the specified disk on each node can be written to and read from. You can choose the standard time period of 1 minute, or you can specify your own value.
• Run this resource in a separate Resource Monitor—You should enable this option so that each GeoCluster Replicated Disk resource runs in its own monitor.

For more information on Advanced Policies options, see your Windows documentation.

5. Connection parameters—This tab controls disk replication, network routing, and orphan files for Double-Take Availability.

• Disk to replicate—The volume to replicate
• Network to route Double-Take mirroring and replication traffic over—The network to use for Double-Take Availability mirroring and replication traffic. If you do not have multiple networks established, you will only be able to select the one network that does exist. If you do not select a network, GeoCluster will use DNS to determine a network route to use.

Ideally, the networks used for various traffic should be separated. This is dependent on the number of networks that you established when you created the cluster and the priority assigned to each network. For example, if you have two network routes, separate GeoCluster and your public traffic. If you have three routes, separate the public traffic and then separate GeoCluster from the cluster heartbeat.

Modifications to either of the first two settings will not take effect until the next time the resource is brought online.

• Interval to check unresponsive nodes—The frequency to determine how often an unresponsive node is checked to see if a Double-Take Availability connection can be made
• Delay connection until resources dependent on this one are online—This option allows you to delay a Double-Take Availability connection until any resources that have the GeoCluster Replicated Disk resource as a dependency are online. By ensuring that all resources that are dependent on the GeoCluster Replicated disk resource are online before starting the connection, the chance of a conflict occurring because application resources are attempting to open files exclusively while GeoCluster is mirroring those files is removed.
• Enable orphans—An orphan is a file that exists in the target location but is not in the source location. You can enable the resource to remove or delete orphans during a mirror. If you choose to move orphan files, specify the location where you want to move them. You should not select the system volume for the orphan files because it could impact the stability of the cluster service.

6. Compression—If you want to configure Double-Take Availability compression, verify that Enable Compression is selected. Depending on the compression algorithms available for your operating system, you may see a slider bar indicating different compression levels. Set the level from minimum to maximum compression to suit your needs.
7. Online Pending—Because context-sensitive, right-click menus are not available in the Windows 2008 Failover Cluster Administrator, GeoCluster processing controls have been added to a properties tab. For details on this tab, see Monitoring a cluster workload.
8. Mirror Properties—This tab controls the Double-Take Availability mirroring process.

• Full Mirror—All files in the replication set will be sent from the source to the target.
• File differences—Only those files that are different based size or date and time will be sent from the source to the target.
• Send data only if Source is newer than Target—Only those files that are newer on the source are sent to the target.

Note:

If you are using a database application, do not use the newer option unless you know for certain you need it. With database applications, it is critical that all files, not just some of them that might be newer, get mirrored.

• Use block checksum—For those files flagged as different, the mirror performs a checksum comparison and only sends those blocks that are different.
• Calculate Replication Set size prior to mirror—Determines the size of the replication set prior to starting the mirror. The mirroring status will update the percentage complete if the replication set size is calculated.