@cpposteve after calibration, the read back should be around +-2 counts (when rounding the output to three decimal places). My unit was in a similar state as yours, its last cal date was in 2014, had ancient firmware, and was off by +-20 counts. And now it finally agrees with my benchtop DMM. Though I now question the effort needed to make that happen.
@Gandalf_Sr all you would need to run it is the Java JRE or JDK, a simple Java IDE like BlueJ (to edit, compile and run) or use Notepad++ to edit and Windows PowerShell to compile and then run it. I could turn it into a jar, but you really need the source code to make tweaks as are inevitably needed. Of course being Telnet based, you would need a wired network with a DHCP server, or assign static addresses, and know what device has what address.
To see what its all about I attached a .txt output of the voltage calibration routines for all three channels. It's still a rough draft and may need a few adjustments to work with another DMM, but that wouldn't be hard to do.
Now if only Rigol could improve their quality control regarding firmware bugs and include all the files you actually need when updating...