On the parent SMP, multiple machines appear as "Untracked". In the Child SMP, those same machines appear as tracked and fine.
The "Untracked" status means that if a computer is not targeted by any "Targeted Agent Settings" policy, then the overall status is counted as 'untracked' (KB 150166 "What the new Agent Health statuses mean?").
These machines were not in the default target needed for the "Client targetted settings" because most of the inventory data was not present on the Parent SMP database.
It is possible that during a recent replication, the machines didn't update their information and since the child SMP already sent those details, it is not trying again to do it.
The simplest solution is to force the necessary information to be sent again from the Child SMP.
Two simple ways to do it:
1. Run the following query on your child SMP database:
delete from ForwardingHistorySummary
where ResourceGuid = 'GUID_of_the_Computer'
and then run a differential replication from the Child to the Parent SMP (just look at Task Scheduler on your child SMP and run it).
After the replication is over, check on the parent SMP and see if that machine now looks right.
2. If the above doesn't work or there are too many to manually clean up, then run a "Complete Replication" from the child SMP to the parent SMP. That should send up everything that we should have for that machine.
Note: By default, complete replication is turned off. So, on the parent SMP, go to Settings>Notification Server>Hierarchy. Right-click on the child server>Edit. Then, under the Schedule tab, just check the box for the Complete schedule. After that, you should see a Complete Replication scheduled task on your child SMP's task Scheduler.
KB 171656 "Computers are showing untracked on console"