How to Fall Back to the Previous Version of the Software Stack (Command Line)

The fallback operation "undoes" the last update attempt of a managed host, whether it finished or did not start.

For example, three managed hosts are updated one at a time. The update completes on the first managed host. Then, the update begins on the second managed host. When the first managed host boots the updated boot environment, you notice that there are problems with the system. You cancel the running update.

Each of the three managed hosts are in a different state. The first managed host completed the update. The second managed host started the update, but did not complete it. The third managed host did not start the update.

The fallback operation ensures that each of these managed hosts reverts to the boot environment running prior to the update attempt.

The fallback feature fails if the system cannot boot. In such cases, see " Solaris Live Upgrade (Overview)" in Solaris 9 Installation Guide.

1. Determine which managed hosts you want to fall back to the previous active boot environment.

   For example, perform the fallback operation on the /web-server/host1 and /web-server/host2 managed hosts.

2. Fall back to the previous version of the software stack.

   $ changemgr fallback [ -u username ] [ -p file ] [ -d domain ] \ topopath ...

   topopath

   Specifies the path to a managed host or host group. The managed host or host group is relative to the top of the specified administrative domain.

   For descriptions of the other options, see How to Add Managed Hosts (Command Line).

Example-Getting Managed Hosts to Fall Back to the Previous Active Boot Environment

Suzi wants the /web-server/host1 and /web-server/host2 managed hosts to fall back to the previous active boot environment.

$ changemgr fallback /web-server/host1 /web-server/host2