When the indexing is done, a runtime is being executed in the guest OS (which is removed after the indexing job). There could be many reasons why it suddenly takes hours. The VM itself is overloaded with doing something or there is an issue with retrieving the information from the OS.
I would suggest to start with looking into event viewer inside the guest VM after it has happened and see if you can't find something special. Maybe you have a monitoring solution to review some performance counters from that VM?
If nothing seems abnormal, you can always create a support call and let our engineers dive into the logs. Maybe they can find something specific on our side. But as said, it is best to start in the VM first
My 2 cents