groupmonitor.lua is sometimes faulty

[rbourgeon]

Hello,
In my company we handle a pretty big pool of LogicMachines and we noticed that sometimes, event-based script that are active, don't trigger when they should. Here are more details about the issue:

• It can happen on LMs with any firmware (we noticed that issue on LMs with 2017 firmwares as well a LMs with 2018, 2020 and 2021 firmwares)
  
• Most of the time, when it happens, the LM has a pretty long uptime (it hasn't rebooted since 50 or even 100 days)
  
• We noticed that by placing a simple "log('hello')" at the beginning of the active event-based script, then when writing into the trigger object, the log doesn't appear in the Logs panel.
  
• When it happens, it doesn't happen for every KNX object of the LM, but only for some of them (which means some event-based scripts keep on triggering when they should, and some others don't).
  
• Since "groupmonitor.lua" is the process responsible for listening to the bus and triggering the event-based scripts, we tried killing groupmonitor.lua and then waiting for it to be reloaded (I think there is a daemon/cron job that runs every minute and launches groupmonitor.lua if it isn't running). 
  
• After reloading groupmonitor.lua, the problem disappears and every event-based script triggers properly.
  

What do you know about this issue? Would it be possible to imagine a script that could detect this issue on a LogicMachine, and when it detects it, kills groupmonitor.lua? Would it be possible to do it without any false positive (we think that the problem is happening while it actually isn't) nor any false negative (the reverse)?