This is a difficult answer, as it may vary a lot depending on your setup. Can you give a better context?
So, you have a cluster of 2 appliances.
What are you using them for? Do they host business critical applications that resolve to F5 IPs? In this case, when both units are down applications will be unavailable, and there's no way to bypass F5 easily since you'll need to route all traffic in another way.
There's external appliances that can act like a "bypass" and monitor links to F5 devices, and when they see links down they forward traffic downstream on another link, but in this case destination will still be F5 IP's so you need to plan for a disaster plan that might involve updating DNS entries.
A workaround to this problem might as well be configuring F5 virtual servers with the same IP's of BE servers, so that with a bypass device there will still be something that respond to those IP's, but again with so little context it's difficult to consider.
Another thing that comes to my mind, which is the most straightforward, will be upgrading your cluster to three devices. If application availability is so critical to you that you need to plan for a double fault, this might as well justify the budget to improve redundancy.