Please forgive my ignorance, I am trying to provide the best information for you to be able to help me. I am limited in my understanding of this Mitel box, and the lack of documentation for me to refer to makes it all that much harder.
The server that gets the SMDR data is a generic PC running Ubuntu and some application called VTLogger. The VTLogger application has been heavily customized for our purposes. This server has a VoiceTronix board installed with a couple of T1 interfaces. I believe that the board is some sort of a Sangoma WanPipe variant. On one of the T1 interfaces, there is a cable that is split from the line that feeds into the Mitel. The VTLogger box also has a connection to the network, and the usual connections to a keyboard, mouse, monitor, and power are all there as well. The VTLogger box does not 'pull' the SMDR data, that much I know for sure. The SMDR data is delivered to it via FTP shortly before midnight every night. I simply assumed that these text files were being put there by the MItel box, but you seem to think otherwise and I trust you know what you are talking about.
There is another piece to this puzzle that is called Prairie Fyre. We are not certain if this is relevant.
Normally, the text files that contain all of the SMDR data are FTPd to the VTLogger shortly before midnight. At exactly midnight, a process runs on the VTLogger that imports all of the SMDR data to a MySQL table and then moves the text file to a different directory in case it is ever needed again. This database import process fails when it tries to process the files that are generated for Saturday and Sunday because the files don't contain any SMDR data and instead look very much like the image what is shown on the image I linked to in my original post above. We just want to figure out why the SMDR is not being captured and it is a bit of a challenge with our limited knowledge of the system.
We do use queues and I think that is what our agents 'plug into' - I am not sure if they are ACD agents.