Server 2012 RDS - User Profile Disks - Errors during Logoff

Question

I have set up a test Server 2012 RDS collection (Single Server for now) and implemented User Profile disks.
I have two problems.
First: My generic test user can connect and does successfully use the user profile disk as expected. However, at
logoff, the system event log contains these errors:
The error (NTFS 137) is: The default transaction resource manager on volume C:\Users\ts3.test encountered a non-retryable error and could not start.  The data contains the error code.
The warning (NTFS 50) that concerns me is:
It appears that the user profile disk is being "disabled" or "disconnected"
before the profile data is completely written at logoff. What can I do to troubleshoot this?
Second:
Update: A post from Mike Connor on the following page: -LINK- solved
the problem described below. 
My administrative user always logs on now with a temporary profile. At the beginning, the UPD was working and mounting. That stopped working. In attempting to troubleshoot, I logged the admin user off and deleted the UPD disk
file from the share. I remember it working again after generating a new UPD disk file in the share. Soon, it quit working again. I deleted the UPD disk file again from the share and ever since, it has never regenerated a new UPD and
always logs on with a temporary profile.

Answer 1

We experience less to zero problems after an outage with our User Profile Disks when we use the option: "Store all user settings and data on the user profile disk". Other then when we use the other checkbox (even though all checkboxes
underneath are checked). 
Our test setup was, an RDSH 2012 r2 server(s), with an 2012r2 server holding the share.
We first tried the "store all user settings" for the User Profile Disk, we logged several accounts onto our RDSH Server we then disconnected our server which holds the share to simulate an outage. After about 5 minutes we connected the
server again holding the share. When the share comes back online, we logged off the users and logged them back in. The result was a valid profile for all the accounts including their last changes from before the share got disconnected.
The exact same test was done with the option: "Store only the following folders on the user profile disk" with all checkboxes underneath checked, which gave us corrupt profiles.. We also did these tests by giving the server holding the share a reboot instead
of an disconnect from network. This gave us the same result. Each test was done at least 2 to 3 times..
I hope others can confirm this as well in their situation, as it seems to be a bug within that specific checkbox.
Looking at the above tests, the best way in my test cases were to exclude the default folders (f.e.: \Documents) if you would like to redirect them instead of using them inside your User Profile Disk. Redirects can be done through GPO Folder redirection
options or GPO registry combined with Environment Variables.
Even after several attemps to simulate an outage, i get left with consistent profiles. I've now also included an hard reset of my virtual machine (that gave the same results as a disconnect of the network card). This was all done through using the: "Store
all user settings in the user profile disk". The other option left us with inconsistent profiles. I hope others can confirm this as well..
Looks to me that the type of connection towards the User profile happens different. You can see this when looking at the C:\users\<username> and subfolders if you use user profile disks. The 2 options show different symbols as well.