This is the result of a known issue. If an HTTP profile is applied to a virtual server deployed in a non-default route domain (a route domain other than 0), the output of the tmsh show sys conn command fails to show the selected pool member and instead reports ::.0. Additionally, all addresses are represented using the route domain internal addressing scheme. This makes it difficult to view or manage connections using the tmsh show sys conn command. For example:

• Without HTTP profile:
  
  2620:0:c10:f501:0:1:a02:1:35546  2620:0:c10:f501:0:1:a02:97:80  2620:0:c10:f501:0:1:a01:1:80
• With HTTP profile:
  
  2620:0:c10:f501:0:1:a02:1:35546  2620:0:c10:f501:0:1:a02:97:80  ::.0

Note: To convert route domain internal addresses to the normal IP%ID notation, refer to the workaround section of SOL10467: Userland applications on a BIG-IP system cannot connect to hosts in non-default route domains.

F5 Product Development is tracking this issue as ID 340468.