Upgrading Active-Standby LTM into v11.5.3

We did a similar upgrade a couple of months ago with one of our customers (8900 pair from 10.2.4 to 11.5.1 HF8). They had some 150+ virtual servers, partitions, etc + much more for 10k active users. All their DMZ DC traffic went through the F5 pair (wildcard virtuals in all vlans).

First we validated the config upgrade upfront in our la lab VE environment to find out any possible config issues in the migration, some of which they had and were taken care of. This was also a prime time to do cleaning of the config from clutter accumulated over several years.

Took away any mirroring, gateway & vlan failsafes and similar, which might cause automatic failovers. Also considered for fast failback scenarios. So a quite bit of careful planning and preparing/testing upfront. All sw was downloaded to units and ready for upgrade 1-2 days before the upgrade. Users were notified of possible outages to service during the upgrade day. A small test team was recruited to test the most critical applications.

After this the upgrade was done successfully within in one half of a weekend day during a service break. HA pair was broken for a few minutes boot times, with one unit active (either 10.2.4 or 11.5.1) all the time. No noticeable disruption to services at all. Especially no problems with any of the passthrough traffic through the wildcard virtual. Think about no or less than 5 sec outage for any service..Originally it was planned to run on 11.5.1 as single for a couple of days and upgrade the other unit in the pair after that, but as everything went very, very smooth, and the testers were very happy, the customer decided to boot he other unit up to 11.5.1 the same day. We had a backup plan, but fortunately not needed.

We had planned for a long weekend for the actual upgrade, but all was done during one Saturday morning..systematic approach, careful planning, backup plan and pre-validating the config upgrade made it a success.