Raspi crashes in while loop

[diy-guy]

Hi all,
i have an problem with a script i wrote.
When running this infinityloop script, my Raspberry Pi 3 freezes after a few minutes. I have to force reboot my pi to get it working again.
This Script checkes if my girlfriend or me is at home with our smartphones and activates or deactivates the HDMI Port to save some energy.
It would be nice to get an hint what causes this and how to fix this problem

Code: Select all

#!/bin/bash 
MACME="12:34:56:0D:FE:10" 
MACHER="12:34:56:3A:4C:54" 

while :  
do 
if sudo l2ping -c 1 $MACME &> /dev/null || sudo l2ping -c 1 $MACHER &> /$
   then
        vcgencmd display_power 1 
        sleep 20 
   else
        vcgencmd display_power 0 #deaktiviert den HDMI Port
        sleep 10
fi
done

[DougieLawson]

I've no idea what you're trying to achieve by running that script that does stuff with vcgencmd periodically, can you explain what it's trying to do?

The first thing you need to do after pulling power (or shorting the RUN header) and rebooting is to look at what diagnostics you've got available. So start by looking at: /var/log/dmesg, /var/log/messages & /var/log/syslog, you're looking for any messages in there that don't appear normal or informational. Is there a kernel oops? Is there something like a USB device driver that's killing your system? Has the kernel out-of-memory (OOM) killer been activated?

Next is to set up a cronjob so that you record the output from the free command every minute as that will tell you if something is leaking storage.

There's no exact science involve here, it's just a case of forensics and hoping you find something that's being triggered that shoots your operating system dead.

Also, remember that vcgencmd is a sample program and, possibly, hasn't been fully debugged by the RPF folks who wrote it.

[diy-guy]

I use vcgencmd because it is easy to use. I can sent an "vcgencmd display_power 1" even if my HDMI Port is running, without having my monitor flickerung as it is with the "tvservice" command.

This script checks if my smartphone is reachable to have some kind of presence detection.
If i am not at home the Monitor is in sleep mode an reduces energy consumption to save some money.
This is an Smart Mirror Project.

[diy-guy]

I just watched those logfiles at the time my errors occured but there is poorly no useful entry which shows me an error.
I also wrote a script which just l2pings my smartphone without this vcgencmd command.

So there seems to be a problem with permanently (5sec delay) pinging an Bluetooth MAC adress via l2ping. This script works for about 10minutes then the screen freezes and my PI3 isn´t responding via ssh and has to be hart rebooted.

[stderr]

I just watched those logfiles at the time my errors occured but there is poorly no useful entry

If whatever is happening locks up the entire machine, would there be any way for it to write a useful entry?

So there seems to be a problem with permanently (5sec delay) pinging an Bluetooth MAC adress via l2ping. This script works for about 10minutes then the screen freezes and my PI3 isn´t responding via ssh and has to be hart rebooted.

No one is going to know if there is some problem unless they happen to have done exactly the same things you are doing. But clearly the answer is to debug your program.

Question number one, what line in your script is causing the problem, if it is really the script at all. Maybe you've got a hardware problem or something else in some other software is messed up.

So debug it by removing pieces that seem most likely to be a problem and see if the problem still occurs. You can't do this so easy if it takes days to crash but your claim is five minutes. Remove the bluetooth pings and see what happens. Does it still blow up?

Let's say you find out it is the bluetooth that kills everything. That's huge because then you can ask a question about bluetooth and not a question about your program, which only you have any experience with. Or better, first google about the topic to see if others are having bluetooth ping lock up problems on the pi, probably have to ask more generally but maybe not.

Leaving aside the logic in trying to find out if you are home every five seconds in order to save money by turning off a monitor. Why not every five minutes? Even if you feel like you need to turn the monitor on more quickly when you get home and can't just wiggle the mouse, at least *test* with a longer delay. You want to start to get a feel for why the system is blowing up when you bluetooth ping, if that's what is happening.

[DougieLawson]

... at least *test* with a longer delay. You want to start to get a feel for why the system is blowing up when you bluetooth ping, if that's what is happening.

This is a good suggestion, with a longer delay does it take longer before it dies? Try with a shorter delay, does it die quicker?

[RaTTuS]

why sudo ping ?

[diy-guy]

thanks for you help.
meanwhile i found out that it is clearly the l2ping that crashes my Pi. If the delay between the ping is longer it just takes longer to freeze my device. I just move the problem to an later point but does not solve it with an delay of 5min.

But this error seems to be prominent and well known since 2002 (here i found the first problems whith freezing devizes while l2ping an MAC adress)

https://groups.google.com/d/msg/openhab ... PNS9IcJhwJ
viewtopic.php?p=410269#p410269

Poorly none of them found an solution for this problem. Maybe an Problem with BlueZ stack?

[Aydan]

See if you can use bluetoothctl scan instead of l2ping.

[DougieLawson]

why sudo ping ?

Because ping as a non-privileged user was badly broken for a while. Latest version is fixed.

[diy-guy]

You should reproduce this error by yourself if you do a simple floodping for a few minutes:

sudo l2ping -f 12:23:34:45:56:67

it takes about 2min until my Raspberry Pi crashes while pinging my iPhone.

you can find your MAC Adress via

hcitool scan (make sure your BT device is in pairing mode)