In relation to this article https://my.f5.com/manage/s/article/K02805156, we resolved the issue by manually configuring the FQDN nodes to pool on the paired device prior to syncing.
Sadly we don’t use FQDN because it comes with some limitations such as not being able to disable individual pool members that are auto-populated from DNS records. Ideally we should be using a higher layer health monitor but in some instances those can’t be used and it makes it significantly more difficult to pull servers out of rotation without disabling a large group. Good find and thank you for publishing.