Typically we upgrade switches one switch at a time, the entire switch all at once. The firmware is downloaded and applied when the switch is reset. Any lag configured across different slots to provide high availability does us no good when performing an upgrade of the switch. That is unless we perform a High-Availability upgrade. A high-availability upgrade performs a rolling upgrade to individual slots. By default, each slot is upgraded one at a time allowing all other slots to remain operational. The key for us is to make sure we design switch connections to take advantage of this capability. Any lag connections we create on the switch should be spread across different slots, and there should be enough members of the lag to allow the slots not being upgraded to pick up the traffic of the one being upgraded.
But before going further, we need to understand the restrictions on when High-Availability Upgrade (HAU) can be used.
- Configuration restore points cannot be set. There must be no change in configuration as a result of the rolling upgrade.
- There must be at least two HAU groups, each one having an operational module.
- HAU compatibility key must match between the current image and the target image, meaning that there are no major feature changes between the two versions.
- S4, S6, and S8 switches require at least two fabric modules.
- Virtual switch bonding cannot be broken during the HAU process. The bonding must exist across two different slots on each chassis. One exception to this is that if all slots in the same chassis are in the same HA group, HAU is permitted if only one slot in each chassis is used for bonding the chassis.
Verify Target Image Compatibility
As stated earlier, the image version cannot contain any major feature upgrades. Enterasys has made it easy for us to determine whether a target image is compatible with the current image by including a HAU compatibility key for each image. You don’t actually have to verify the key itself. If the key is appended with “HAU compatible”, that means the image is able to be used as an HAU for the current image.
To view the current boot image, use set boot system.
To view all copied images and HAU key, use dir images. Notice the HAU Compatible next to the HAU key for image S-81104-0005.
Configure Boot Image and HAU Mode
The initial setting for HAU mode is never. If you want to perform a HAU, you must either change the default HAU mode to if possible (do HAU if possible, otherwise do standard upgrade) or always (do HAU if possible, otherwise do no upgrade) or use the system boot command to perform a HAU. Enterasys recommends not changing the default HAU mode unless about to perform an upgrade, then change it back to never. I don’t bother changing the default mode and just use system boot mode high-availability instead.
Configure HAU groups
By default, each slot is assigned to its own high-availability group. This is the best way to spread out the upgrade across the minimum number of slots at a time, but also increases the total upgrade time. It may be desirable to group together slots that do not share lags.
To view the HAU groups use the show boot high-availability command.
To assign slots to HAU groups, use the command set boot high-availability group <group-#> <slots>. The example below sets slot 1 and 2 as members of HAU group 1.
Configure HAU delay
There is about a 5 second delay following the upgrade of one HAU group and the beginning of the upgrade of the next HAU group. It is possible that not all protocols will be up by then, so an additional delay can be configured. This delay can be up to 600 seconds.
To configure an additional delay between HAU groups use the set boot high-availability delay command. This example sets the delay to 25 seconds.
If the boot high-availability default mode is set to if-possible or never, you can simply use the set boot system <filename> command to initiate the HAU. I would use the high-availability option along with this command and not rely on the default mode being configured to do a HAU.
Force Early Completion of HAU
If the HAU is taking too long and you want to complete the process immediately, you can force the upgrade to complete an upgrade of remaining HAU groups at the same time.
To force the early completion of a HAU
This is an example of the upgrade process. This example includes verifying HAU compatibility, setting the delay, verify default group memberships, and initiating the upgrade. It also includes verifying the current boot image.
That is the complete process of performing a High-Availability Upgrade. It is a relatively simple process and very helpful. The important thing is to plan your LAG and bonding connections to take advantage of this capability.