A problem seemed to occur when the channels related to the ASC model were read in the 2nd subset of 12 channels.
So reading 12 channels in simultaneously is probably not a problem.
After reading the 1st subset of 12 channels, garbage collection or port closing doesn't work...?

Check list for the next maintenance day.

1. Read 20-30 channels from other 2kHz model on ASC computer such as ADS model (it's problem only on ASC model or not).
2. Read 20-30 channels from other 2kHz model on other computer such as VIS TWR models (it's problem only on ASC computer or not).
3. Read 20-30 channels from 16kHz model such as LSC model (it's problem only on 2kHz model or not).
4. Read 20-30 DQ channels instead of TP from ASC model (it's problem only for TP or not. Though sampling rate is different between DQ and TP, sampling rate is not a issue if 3. is passed).
5. Read 20-30 channels by multiple process per 12-channels as follows (it's problem on garbage collection of the script or not).
   #!/bin/bash
   for x in ...
   do
       python3 ./subtract_dark_offset.py [12-channels]
   done
6. Measure 12 channels from ASC model and then measure another 12 channels from ASC model in one diaggui process (It's problem only on cdsutils or not).

These are only what I can think of right now.