For the ATLAS Phase-I LArC carrier + LATOME AMC cards we’re exploring using the CERN IPMC. The system features are implements and basically work, but we have a system instability. If no LATOMEs are plugged in, the system is stable for as long as we’ve run tests (>1 week). When we plug in LATOME’s, however, the LArC and LATOMEs reboot after some hours, although the time varies. We have simple uptime counters on both the LArC and LATOME boards and they show an essentially simultaneous reboot. We have started using SOL for monitoring, but don’t have results yet.
Our question is whether there is a way to (temporarily) adjust the watchdog timer on the IPMC and/or to get a console error message if the watchdog times out to see if the reboot is being caused by processing time. Is this possible?
there should be a message on the IPMC serial debug interface UART if the watchdog timer expires and the IPMC reboots. It is not possible to change the watchdog timeout at runtime, this value is predefined and set to 2 seconds, which is already very long. We have never actually observed the IPMC rebooting because of a watchdog timeout, however this could potentially happen if you have user code (e.g. OEM commands or similar) which use an excessive amount of CPU cycles.
Thanks. We shouldn’t be anywhere near 2s unless there is a hang up on the I2C that holds the CPU. We’ll look for the watchdog message in our SOL sessions.