process 'reactorra' taking an enormous amount of CPU

[rbourgeon]

Hello,
The CPU of a LM is very overloaded (load average ~ 3.00). We decided to investigate it.

When we run the command `top`, we see that the process responsible for managing the inputs/outputs (`lua /lib/genohm-scada/plugins/reactorra/daemon.lua`) eats up a good part of the CPU resources: column "%CPU" indicates a number that is almost all the time around 30% for this process, but we've seen it go as high as 57%!

This is very strange, given that:
- There is no input/output connected to this LM (all 16 rows in the "Reactor" tab are blank in the scada-main interface)
- This LM is equipped with the latest firmware (20251204).

Model is  LM5 Reactor IO V2/V3 (i.MX6).

Do you have any idea where this could come from and how we could solve it?

[admin]

Have you tried rebooting LM, does it help? Are there any other processes with high CPU usage?

[rbourgeon]

Hello, yes when we reboot the LM it solves the problem but only temporarily. No other processes have a high CPU usage.

[admin]

Do you have anything in LM Error log? Does this issue happen on one LM or multiple? If you have another Reactor try swapping the SD cards between them and see if this issue happens on the second device.

[florentd]

Hello,

Just following up on rbourgeon’s posts with a few more details from our side

We’ve now seen the same behaviour on several LMs. As far as we can tell, it only started after we updated to firmware 20251204; we hadn’t encountered it before that

When it happens, there’s nothing in the LM error log
We haven’t tried swapping SD cards yet, as the affected devices are installed at customer sites

So far, we’ve tried:
- re-applying the firmware update
- removing the reactorra package — CPU usage drops — and then reinstalling the package
- rebooting the LM

All of these fix the issue temporarily, but the reactorra process eventually starts using excessive CPU again, for no obvious reason. (Note that we think re-applying the firmware only works because it triggers a reboot)

A reboot works temporarily, but it isn’t really a sustainable solution. Is there anything we should check or capture next time it happens, before rebooting?

[admin]

Can you send serial numbers of these LMs via PM?