since few week I'm observing a strange behaviour of my linux box. It randomly restarts without any evident cause. messages, syslog and dmesg do not point to any direction. Looking at bios (dmidecode) neither brings no news. I also redirected kernel messages at grub level to a serial port, another pc is logging serial output. In spite I had already two reboots I didn't get any trace of the cause.

can a defective PSU cause such issue without leave any track? can anybody sugest me any other way to investigate on this matter?

  • This server is composed by an Intel Atom D525 on an Intel D525MW mainboard with firmware version MWPNT10N.86A.0083.2011.0524.1600. There is the PCI slot populated with a Sangoma A200 card. Just one USB peripheral: a ps/2 to USB adapter. PSU is 150W – alessandro Mar 17 '14 at 15:52
  • PSU might be the cause or memory, or few other options. Try to load it with some CPU-heavy or I/O heavy task. See if it would restart under load. – Dmitri Chubarov Mar 17 '14 at 15:56
  • If the cause is memory, which I tested twice, shouldn't I find anything on kernel messages? – alessandro Mar 17 '14 at 16:03
  • I want to apply your suggestion: heavy load on CPU and I/O. Could you please hint me how to do? – alessandro Mar 17 '14 at 16:05
  • Anything. Computing md5sum of your filesystem or dd if=/dev/urandom of=/dev/null for instance. Just to see if higher load would cause a reboot. – Dmitri Chubarov Mar 17 '14 at 16:13

If there are no indication in the logs then it's hardware related, typically motherboard, memory, CPU. For memory the indicated read/write tests with 'dd' can be used, since that's fine you can look into CPU temperature (BIOS), otherwise it can be 'anywhere' in the motherboard. Another issue could be the power supply and incoming brown-outs/spikes.

  • Yea, I'll second this opinion. Almost certainly hardware, could be almost anything: faulty motherboard, etc. – Michael Martinez Mar 19 '14 at 17:59
  • As I wrote in my las post, I found evidences problem lays in faulty PSU, infact, After unassembled it, I notice it is a poorly assembled piece of hardware, with two capacitor poorly soldered. However just for my understanding, let's assume bad memory with some faulty locations. the worse it can happen is some memory fault, and shouldn't it be seen on kernel logs – alessandro Mar 20 '14 at 13:14
  • The faulty PSU seems to be the most probable cause, but there still something I can not understand, and it is: why this problem appeared after months and not as soon as this server started to work? – alessandro Mar 20 '14 at 13:21

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.