ADP Debugging Guide

Log files

/home/adp/log/adp-control.log

This is the main log output of the adp-control service; however, note that raw print statements and uncaught exceptions will be logged to /var/log/upstart/adp-control.log instead, so be sure to check there when debugging.

/var/log/upstart/adp-control.log (requires sudo to read)

This is the raw log output of the adp-control service. It should only contain print statements, uncaught exceptions and upstart errors; all other messages will be logged to /home/adp/log/adp-control.log.

Quick initialisation

To reinitialize ADP quickly, the INI command can be given the parameter "NOREPROGRAM" to skip reprogramming of the FPGAs. If the servers are already powered up, the resulting initialisation takes only about 10 seconds. A SHT command is not necessary for simple reinitialization.

Talking to adp-control from the command line

The script:

/home/adp/lwa_sv/scripts/test_adp_control.py

can be used to send queries and commands to the adp-control service. With no arguments it will send a short sequence of test messages. Other ways to use it are:

$ ./test_adp_control.py status

$ ./test_adp_control.py INI

$ ./test_adp_control.py INI NOREPROGRAM

$ ./test_adp_control.py SHT

Checking data flow

The script:

$ /home/adp/monitor_p5p1.py

can be used to check that data packets from the roaches are being received on a server. It prints out rates in units of seconds, averaging over 2 second intervals.

Data capture within a running ADP server pipeline can be monitored using the following script, which will give a detailed summary of current capture statistics:

$ /home/adp/bifrost/scripts/print_Depacketize_stats.py adp1:7777

Common problems

Forgetting to update adp-control scripts in /usr/local/bin/adp

Forgetting to adp_config.json in /usr/local/share/adp/

To restart the adp-control service:

$ sudo start/stop/restart adp-control