I am preparing to upgrade to CATOS on a 6509 with redundant SUP720 from 8.3.3 to 8.6.6a. I'm comfortable with the upgrade process itself, but the Cisco documentation states that "Note: There is no software image version compatibility in the 8.x software release train. This includes major releases, such as 8.1(x) to 8.2(x) to 8.3(x) and so on. This also includes sub-releases, such as 8.1(1) to 8.1(2), 8.2(1) to 8.2(2) and so on."
This same document states the following:
"Note: If two software images are incompatible, the software upgrade process impacts the system operation (that is, be greater than the one to three second switchover time of a high-availability switchover) and no NVRAM configuration changes are synchronized between Supervisor Engines."
The document in question is found here:
Does anyone have experience with the amount of time I might expect for high-availability switchover when my images are not compatible?
Thanks for your brainpower...
the same document in the Hybrid mode section after explaining that fast switchover is default redundancy strategy until 8.5 says about fast switchover:
>> The switchover performance with default settings that results takes approximately 28 seconds, plus the time it takes for the protocols to restart. High availability removes this limitation. High availability allows the active Supervisor Engine to communicate with the standby Supervisor Engine. This keeps feature protocol states synchronized. Synchronization between the Supervisor Engines allows the standby Supervisor Engine to take over in the event of a failure.
>> By default on the Cisco Catalyst 6500 Series Switch, the Catalyst OS software images on the active and the standby Supervisor Engines must be the same. If the two supervisor images are not the same version during system bootup, the active Supervisor Engine downloads its current boot image to the standby Supervisor Engine. The NVRAM configuration of the active Supervisor Engine is also synchronized between the Supervisor Engines.
high availability as IOS SSO requires identical images. Now it should be understood if during transition the system falls back to fast switchover strategy.
I would say it takes the time to reload both supervisors so it has an impact and you need to ask for a maintanance window.
Hope to help