Discussion

Bananian shuts down after a couple of days

12 4477
koev36  
Hi,

I am using bananian 14.8. Unfortunately the banana pi shuts down every couple of days (3-4 days?). At that time no LED is on anymore.

I looked at the messages and syslog but didn't see anything which points me into any direction.

Cheers for your ideas ;)
No come across this, check your power supply

Hmm. Oddly enough, i didnt have this issue until today. I have bananian 14.09 running 24/7. But today after i got home from work, the device was off, no LEDs on and no power to the USB devices (usually when i shut down the Banana Pi I can see the USB devices still get power).

Also nothing in the logs and nothing that suggests there would have been a power outage at home. I will have to keep an eye on that and see if it happens again.

tkaiser  
no power to the USB devices (usually when i shut down the Banana Pi I can see the USB devices still get power)


You should use a monitoring solution to permanently check the 'health' of both SoC and PMU (since when there's an issue with insufficient power supply or too much load/current then the device will shut off and nothing gets written to any logs unless you setup something prior to that).

RPi-Monitor for example can be easily installed and configured to use the totally different thermal as well as PMU probes compared to the Raspberry Pi. It writes all the 'sensor' stuff to RRDs every 10 seconds so you might get a clue what happened before a hard shutdown even when there's nothing written to log files.

koev36  
Hi,

nice hint thanks. I did install it but unfortunately the voltage is not shown as I pointed out in http://forum.lemaker.org/thread- ... urrent_voltage.html

Do you have a solution for that?

By the way I had another shutdown but I didn't see any peaks in CPU usage or anything

tkaiser  
You should better have a look at actual current and temperatures. Compare with http://forum.lemaker.org/forum.p ... 2&fromuid=33332 for details

Ok, same issue today. I checked the power supply and noticed a strange noise coming from it. So, for the moment, I exchanged it to see if this solves it. Luckily, I have another 5V/2A USB power supply at home (from my tablet). I'll try to setup monitoring this weekend.

tkaiser  
Edited by tkaiser at Sat Oct 25, 2014 05:04

By simulating an 'enclosure with broken thermal design' I managed to provoke an emergency shutdown of my Banana Pi

Since the method to get the CPU's temperature (for details have a look at the explanation) relies on messages being written to syslog this functionality acts also as a simple 'heartbeat indicator'. This is the last line in /var/log/syslog so the next few seconds the Banana Pi did an emergency shutdown:
  1. Oct 24 12:46:56 bananas kernel: [64833.527149] 0x7f4
Copy the Code
Temperature graphs:


Current/Voltage:


Maybe the voltage dropped below 4.6V, maybe CPU or PMU decided to quit due to temperature? Who knows...

IMO there are two lessons to learn: a) use some sort of monitoring (eg. RPi-Monitor, OpenNMS or OMD) if you rely on server functionality and b) keep heat dissipation always in mind when load might increase (operating the Banana Pi vertically and allowing some airflow seems to be enough)

tkaiser, what is the logging intervall that you use?

I set up a simple script now that logs voltage, current, soc and pmu temperature and load to a csv file every minute.

I don't think that the temperature or load were the problem in my case though. At both times my BPi shut down, there were no clients connected to the server and no cronjobs executed, so the load was virtually zero. My best guess so far is the faulty power supply (the new power supply works fine so far), but I will see how things turn out over the next days.

tkaiser  
Edited by tkaiser at Sat Oct 25, 2014 05:04
tkaiser, what is the logging intervall that you use?


I let my daemon measure the temperatures every 7.5 seconds and RPi-Monitor reads the stuff every 10 seconds and writes it to its RRDs.

I don't think that the temperature or load were the problem in my case though.


Yes, in your case it looks like insufficient power supply which causes the PMU to completely shutdown.

BTW: I tested again to get a clue whether my shutdown was related to temperature or maybe insufficient current. So I put the Banana Pi into vertical position (convection) and tested again 2 hours:




I believe it's the PMU's temperature (reaching 57° in the 'broken enclosure' simulation leading to an 'emergency shutdown' vs. 40° when using convection)

You have to log in before you can reply Login | Sign Up

Points Rules