The 5 second delay introduced by Homin to the end of data collection to flush the buffers does not seem to have worked.
So, I tried something different with data collection to see if it would fix our data collection issues. I modified the rx and tx scripts.
Modifications made
1. To tx script – only enable or disable transmission
2. To rx script – In addition to creating a file for data collection, also send all the spead signals.
I enabled the transmission for both pols. The first dataset was collected for about 75 minutes without any issues. I closed the files and sockets WITHOUT disabling transmission from the correlator. The processed output data had no issues.
The second data set collection went ok for about 75 minutes, when X pol errored out and never recovered. Only a reload fixed it.
I am quite surprised that data collection which went smoothly for about 75 minutes would completely crash, and never recover. This points to something other than buffer issues in the switch, I think.