modbus polling and cpu overcharging

[gdimaria]

hi,
I have a LM whose CPU is often overloaded (it even restarts!).
I have analyzed every possible cause, including infinite loop scripts, but nothing. 
The only thing I notice is a continuous telegram traffic from an external modbus gateway to which several devices are connected. 
There are several telegrams frequently updated and/or repeated. 
Could this be the cause of the overload?
Is it possible to monitor the CPU load trend to determine when it overloads?

Thanks in advance

Peppe

[admin]

Install System load app and check which process is consuming max CPU/RAM.

[gdimaria]

Install System load app and check which process is consuming max CPU/RAM.

I already done, those ones are always in pole position... enspecially nginx (there are several users simultaneously pointing their browser to LM).

[Daniel]

The CPU is nice and low. You should wait till the high CPU happens.

[gdimaria]

The CPU is nice and low. You should wait till the high CPU happens.

It's hard to take a screenshot when it peaks. However, nginx is always very high in that case.. as the signed others.

Often CUP/IO gets 3.50 and more... until it reboots!

[AEK]

The CPU is nice and low. You should wait till the high CPU happens.

It's hard to take a screenshot when it peaks. However, nginx is always very high in that case.. as the signed others.

Often CUP/IO gets 3.50 and more... until it reboots!

It seems like you have loop in scripts or somthing like that. you have very high knx tp load/
make objects sort order by update time and see what GA often updates

[Daniel]

Send me your backup to PM.

[gdimaria]

Send me your backup to PM.

 I can't find the attachement button anymore in PM interface...

[Daniel]

It was never there, use some cloud storage and send me a link.

[gdimaria]

It was never there, use some cloud storage and send me a link.

Referring to our PM:



I meant It would be nice to be able to associate a virtual object to the cpu/memory load, as is already possible with the cpu temperature. In order to be able to trend and thus understand the exact moments in which it overloads

[Daniel]

This is KNX TP UART chip temperature not the CPU itself.
https://forum.logicmachine.net/showthrea...8#pid13948