Page 1 of 1

Early RPi 2 randomly segmentation faults process

Posted: Tue Dec 22, 2015 8:13 am
by bullen
The RPi 2 I bought directly after release from pimoroni has started behaving weirdly.

It has black camera and screen latches instead of white.

Anyone had the same experience?

My other 4x RPi 2 (white latches) are 100% stable.

Re: Early RPi 2 randomly segmentation faults process

Posted: Tue Dec 22, 2015 8:34 am
by DirkS
Define 'weirdly'...

Re: Early RPi 2 randomly segmentation faults process

Posted: Tue Dec 22, 2015 8:48 am
by RaTTuS
what does

It has black camera and screen latches instead of white.
mean

can you upload a video to you tube to explain better

are all the cables plugged in fully etc ...
when did iit stop working proper , what had you done

Re: Early RPi 2 randomly segmentation faults process

Posted: Tue Dec 22, 2015 9:39 am
by AndyD
RaTTuS wrote:what does

It has black camera and screen latches instead of white.
mean
This is a B2 with black latches (from Wikipedia).
This is a B2 with white latches (from raspberrypi.org)

Re: Early RPi 2 randomly segmentation faults process

Posted: Tue Dec 22, 2015 9:40 am
by bullen
So the white latches are just to point out it's one of the first RPi 2 out of the factory. And not the same factory as todays RPi 2, the metal on the HDMI port is different too (matte aluminium instead of dark shiny). Edit: Precisely AndyD.

I have changed nothing, it has been running flawlessly for 1 year 24/7 until two weeks ago when it had one core running 100% on a strangely simple stack trace place, I shrugged it off as my fault. But then yesterday and today it has crashed the process with segmentation fault on other seemingly common stack trace places.

I'm suspecting something broke with the RAM when I had power outages a month ago, 3 in a row.

Re: Early RPi 2 randomly segmentation faults process

Posted: Tue Dec 22, 2015 9:45 am
by AndyD
The differences in latch colour and HDMI port etc may be to do with the Country of manufacture.

Re: Early RPi 2 randomly segmentation faults process

Posted: Tue Dec 22, 2015 10:03 am
by bullen
Exactly, so I'm trying to figure out if it's a one time freak occurrence or if others with early RPi2 from the white black latch factory are seeing failures too after 1 year of 24/7 operations.

Re: Early RPi 2 randomly segmentation faults process

Posted: Tue Dec 22, 2015 10:32 am
by Heater
I'm a white latch kind of guy. Never seen a problem with random failure cropping up.

Have you been keeping your OS updated? Things will have been changing, perhaps introducing a problem.

Have you tried swapping the power supply, just to eliminate that as a possible cause?

Does this problem only show up when running some particular program?

What actually are the stack traces ?

Re: Early RPi 2 randomly segmentation faults process

Posted: Tue Dec 22, 2015 10:54 am
by bullen
I meant black, white latches are 100% stable for me too, typo above...

I have not changed anything.

Power supply is the same as the white latches:

Image

The issue stack traces are not in my code but in Java core classes, in different places every time.

I'm pretty sure the hardware on this RPi 2 is failing.

Re: Early RPi 2 randomly segmentation faults process

Posted: Tue Dec 22, 2015 11:30 am
by RaTTuS
dunno - try a fresh raspbain
then try
sudo apt-get memtester
sudo memtester

Re: Early RPi 2 randomly segmentation faults process

Posted: Tue Dec 22, 2015 12:05 pm
by Heater
bullen,
The issue stack traces are not in my code but in Java core classes...
Ah, now I see your problem. ..... Only joking.

That is a neat cluster you have there.

I'd certainly want to be testing with memtester and such as suggested. It'd start with a new SD card and new OS image just to eliminate those as possible sources of trouble.

Re: Early RPi 2 randomly segmentation faults process

Posted: Tue Dec 22, 2015 12:15 pm
by gregeric
Might it be worth underclocking it a wee bit to see if stability returns?

Re: Early RPi 2 randomly segmentation faults process

Posted: Tue Dec 22, 2015 12:26 pm
by RaTTuS
gregeric wrote:Might it be worth underclocking it a wee bit to see if stability returns?
no
first thing is
viewtopic.php?f=63&t=129721#p866727
anything else and you cannot be certain

Re: Early RPi 2 randomly segmentation faults process

Posted: Tue Dec 22, 2015 2:59 pm
by bullen
Hm, so it's not memory:

Code: Select all

pagesize is 4096
pagesizemask is 0xfffff000
want 900MB (943718400 bytes)
got  900MB (943718400 bytes), trying mlock ...locked.
Loop 1/1:
  Stuck Address       : ok
  Random Value        : ok
  Compare XOR         : ok
  Compare SUB         : ok
  Compare MUL         : ok
  Compare DIV         : ok
  Compare OR          : ok
  Compare AND         : ok
  Sequential Increment: ok
  Solid Bits          : ok
  Block Sequential    : ok
  Checkerboard        : ok
  Bit Spread          : ok
  Bit Flip            : ok
  Walking Ones        : ok
  Walking Zeroes      : ok
  8-bit Writes        : ok
  16-bit Writes       : ok

Done.
Maybe it's the SD card? But I think I'll just buy a new RPi 2 and a new SD card and cross match them if it fails again...

Well see...

It must be the CPU then... how does one verify that?

Re: Early RPi 2 randomly segmentation faults process

Posted: Tue Dec 22, 2015 3:46 pm
by Heater
By all means get a new Pi and SD to get things working.

It would be nice to get to the root cause of the problem though.

I'm betting on, SD card, OS image, power supply...(oh and Java :) )

You could always send me that suspect Pi. For analysis you understand...

Re: Early RPi 2 randomly segmentation faults process

Posted: Wed Dec 23, 2015 6:49 pm
by bullen
Just to follow up now I'm suspecting it is the SD card after all.

When I had the power outages (not good for SD cards BTW) the following script was not run:

swapoff -a
sysctl -w vm.swappiness=0

Basically maybe the swap partition has taken a hit after being power cycled in use after the first failure?

And maybe that's where the problems sneak in... easy way to test is to turn off swap and then the system should work...

But better is to buy a new SD card and make sure!