LM reboots

[gdimaria]

This morning the LM restarted 8 times at a distance of 2/3 minutes from each other. 
I know this because I have inserted a init script that sends an email. 
However, the system only recorded  in the Alert the last "system start" reboot, why didn't it log in all the reboots? 
And what can be the cause of reboots apart from CPU overload?

BR

Peppe

[Daniel]

Anything what happens during first 30 min after reboot will not be saved if reboot happens again. This is how often data is saved on the sd card. Reboot is most likely memory issue.

[gdimaria]

Anything what happens during first 30 min after reboot will not be saved if reboot happens again. This is how often data is saved on the sd card. Reboot is most likely memory issue.

nginx is always on the top of memory usage, maybe because a lot of people are logged at same time.

is it possible to force writing on sd card by a script? I could put it in init script....

[Daniel]

The 30 min rule is for a reason, often saving will shorten lifespan of SD and brake LM. How many users at the same time?

[gdimaria]

The 30 min rule is for a reason, often saving will shorten lifespan of SD and brake LM.  How many users at the same time?

sometimes could be 10 or more users...and I guess they have more than a browser card open at time.

about the 30 min rule, I can understand. But I would run it only in case of reboot.

[Erwin van der Zwart]

Taking away a symptom without addressing the origin / root cause / putting a bandage in place is a bad approach..

You should look into the reasons of the restarts as the controller should run 24/7 without the need of extra savings to SD what certainly leads to a short lifetime span..

[gdimaria]

Taking away a symptom without addressing the origin / root cause / putting a bandage in place is a bad approach..

You should look into the reasons of the restarts as the controller should run 24/7 without the need of extra savings to SD what certainly leads to a short lifetime span..

I agree with you, but that is long time issue without solution:

https://forum.logicmachine.net/showthrea...8#pid37448

I tried everything to understand the reason for these overloads.
 I also trended the cpu and memory load, but it all seems absolutely random.

[Daniel]

If this graph is correct then your CPU is very high.

[gdimaria]

If this graph is correct then your CPU is very high.

yes, that is...

Code:

-- Indirizzo oggetto virtuale
local ga = '50/0/4'

-- Legge il primo valore di load average da /proc/loadavg
local loadavg_raw = io.open('/proc/loadavg', 'r'):read('*a')
local load_1min = tonumber(loadavg_raw:match('(%d+%.?%d*)'))

-- Scrive il valore nell'oggetto virtuale, se valido
if load_1min then
  grp.write(ga, load_1min, dt.float16)
end

 the way I get the cpu load

[gdimaria]

Reboots lately are more and more frequent, and it is not a temperature problem because the CED is air-conditioned. 

After trying to optimize the execution of the multiple scripts, I just have to think that the number of users who connect at the same time also has an influence. 

Is there no way to limit user access at the same time, or at least to warn you that you are reaching the limit? 

I see from the "system load" that the nginx process is often on top of cpu/memory consumption.

Thanks for you attention


Peppe

[Daniel]

If one LM cannot handle it then the only way is to divide project into two.

[AEK]

Reboots lately are more and more frequent, and it is not a temperature problem because the CED is air-conditioned. 

After trying to optimize the execution of the multiple scripts, I just have to think that the number of users who connect at the same time also has an influence. 

Is there no way to limit user access at the same time, or at least to warn you that you are reaching the limit? 

I see from the "system load" that the nginx process is often on top of cpu/memory consumption.

Thanks for you attention


Peppe

share KNX statistic screen, maybe you have loop between event scripts
maybe system load app could show you the problem