Strange thing is that on our Storefront servers (server group) in our central datacenter, the issue does not occur.
These 2 storefront servers are used for internal and external use.
The issue occurs on our storefront servers in the local datacenters of the mills.
Only one storefront server is deployed here for internal use.
The session settings on the central and local storefront servers are the same.
The central storefront servers are loadbalanced on our internal loadbalancer (not netscaler).