Creating, Managing, and Using Account Images

After you migrate your data and applications, you can create master images for each disk in the server through a process called snapshot. The result of the snapshot process is a stable image on disk from which a server can successfully boot and run. The snapshot, or a copy of a server disk in an active farm, is added into an account-level library of images. You can create an account image by using the snapshot mechanism for any combination of software and data present on a server disk.

If you choose, a disk snapshot request can automatically shut down a server to ensure that the resulting image is a stable, production-ready replication of the original image. During the last step of a snapshot request, the server is started up again automatically. Alternatively, you can manually shut down a server to take a snapshot, or take a snapshot without shutting down a server. If you snapshot a local disk, the server is shutdown by the system.

If you want to issue a farm update request after taking a snapshot, first ensure that the server has successfully restarted before you issue a farm update request. Otherwise, the farm update fails. To ensure that the server has restarted, execute the ping ip_address command.

Account images can be created by using the command-line or by using the Image Wizard. See Chapter 3, Managing Software Images in N1 Provisioning Server 3.1, Blades Edition, System Administration Guide. Account images can also be created by using the snapshot tool in the Control Center. See Snapshot Best Practices.

Another type of image, called a global image, can be used by any farm in any account. However, when you use the snapshot tool to create a software image, the image is the result of a snapshot of a disk in use within a farm and is therefore account specific.

For more information on managing global images, see Chapter 3, Managing Software Images in N1 Provisioning Server 3.1, Blades Edition, System Administration Guide.

In addition to software images, the Control Center enables you to take a snapshot of the configuration of load balancers. The snapshot is for element failover purposes. The snapshot images cannot be used to copy configurations to other load balancers.

If you choose to allow the software to automatically shut down the server to take a snapshot of an image, change an image, or delete a disk of an Active farm server, the following process occurs:

• Server is shut down

• Snapshot image is created

• Server is started

This process normally takes 25 to 35 minutes for a 750 MB volume for local storage.

The software enables you to decide whether the server should be shutdown during a snapshot. If you choose to let the software perform a shutdown during a snapshot, the server is removed from monitoring and shut down. A snapshot of the specified disk is made, the server is rebooted, and reregistered the server with monitoring.

However, if you do not want to let the software do the shutdown, the software just takes the snapshot of the disk. The software does not perform any steps related to server shutdown.

For a local disk snapshot, you are not provided with any options. The server is always shutdown automatically.

If you shut down the server manually, a failed device request is generated if the server is still being monitored. To avoid this failed request, before the server is shutdown, unregister the server from monitoring and subsequently register the server on the server's command-line interface.

Run the following command on the server to unregister the server from monitoring:

opt/terraspring/sbin/tsprmonitor -stop \
[ Minutes to reboot (default 20 minutes) ] [ -c]

Running this tool on the server stops monitoring while you shut down the server for the specified amount of time in minutes. The default is 20 minutes. You can run this tool again to extend the time as required. To start the monitoring process again, reboot the server.

The -c option gives you the option to wait for a confirmation that the request to stop monitoring is confirmed. If you use the -c option, tsprmonitor returns with the confirmation that the server is not be monitored. Without the option, you should wait for ten minutes before rebooting the server.