ConfigMgr / SCCM - Package content issues when device is moved from office to remote location

We have experienced an issue affecting some devices whereby they suddenly stop pulling package content down and then not ever running the package install. It doesn't affect all devices, those that seem affected have all had one common issue. On their IPv6 adapter, the "register this connection's address in DNS" was unchecked. Those unaffected have this checked.

Why that's relevant is because these are laptops that will have been on our corporate IPv4 network and then moved to a remote or home location and connecting to the corporate network using DirectAccess which presents an IPv6 address. Why a device still registered with a corporate IPv4 address in DNS should impact what is primarily an SCCM and client pull process is a mystery. Perhaps someone can offer some thoughts.

The observations from such a device that has undergone that scenario. Note: A single content location so no location/boundary issues.

1. Deploy an application package to device
2. Force a policy refresh on a device and it picks up the policy - policyagent.log
3. Execmgr.log reports downloading content and last report is waiting content. Nothing appears in this log after that.
4. CAS.log reports the download has started,last relevant event is "Location update from CTM for content"
5. ContentTransferManager.log reports "CCM_DOWNLOADSTATUS_PREPARING_DOWNLOAD". That's the last event with that GUID.
6. In DataTransferService.log there seems to be a corresponding DTS job which started "DownloadingData" and then "completed : SUCCESS" in 20 seconds.
7. In the ccmcache folder there is a "j.BCWork" folder and a "j" cache folder. The latter is empty. On a good device the BITxxx.tmp files build in the cache until completed.
8. Nothing happens. I can see a number of "BCWork" folders and empty equivalent cache folders that are all time-stamped from when the device was moved to a home location. There are completed cache folders that pre-date that.
9. I have tried a new package deployment and that exhibits the same issue. It looks as though this has got stuck/backed-up and won't process anything. Tried restarting the Agent Host and restarting with no effect.

Any suggestions would be very welcome.

Note: this process has worked and devices that have registered correctly in DNS with their IPv6 address seem ok. Someone did try reinstalling the ConfigMgr client on an affected device and that brought it back to life.