Cisco IOS XE devices boot into a Linux kernel first, then load IOS as a module. If you just power off the device (as we are used to doing with IOS devices), you will see disk-errors (assuming you are connected and monitoring the console) when you power it up that get auto-corrected (hopefully). This happens because log files related to the Linux kernel are still in use when you power off the device.
 
To avoid this, the documentation states to issue a reload before powering down to ensure all the log files are closed correctly, but it isn't clear at what point you can then power off. Of course if you dont, it come-back up as a result of the reload command.
 
I found a link online that recommends issuing the 'reload pause' command instead. When the device gets to the pause, it will show you a 'Enter [continue]…' prompt. At this point, you can safely power off the device and it will not have any disk errors when it boots up again.
 
This assumes you are connected to the console. Not a bad assumption as it is a bit hard to physically power down a router or switch remotely. But if you are not on the console (maybe you have a customer that will to pull the plug for you), you can still issue the reload pause command and wait about 60 seconds. That should be enough time for the device to get to that pause.