0

My desktop had developed a strange habit this morning: shortly after starting up and logging in, it shutdown very suddenly. There was a mechanical noise like a relay opening, the fan spun down, though the power light remained on. I restarted and the same thing happened again. It takes anywhere from seconds to minutes for this to happen. When I went into the BIOS settings I got a message saying that they had been reset. The most perplexing part is that I don't think I changed anything recently that could explain this.

Clearly I'm not hoping that someone will diagnose the problem from these paltry details. But I'm wondering how I can go about collecting more information about the shutdown. I'm running Debian Buster.

I'm really quite clueless here - what (if any) 'blackbox' recording do I have available after something like this?

2 Answers 2

0

You can check the system logs from the time when the shutdown happened. Depending on how sudden and critical the error was, you may not find something, but worth to give it a shot.

If the rsyslog daemon is running, you should find the pain text logs in /var/log/syslog, kernel logs additionally in /var/log/kern.log and other more service specific log files in the same directory.

If rsyslog is not used, check whether the directory /var/log/journal exists, which contain the persistent binary format system logs directly from systemd-journald, the primary syslog daemon on Debian. If it does not exist, these logs are kept in RAM only and are hence lost after the crash. But you can enable persistent system logs simply by creating that directory:

sudo mkdir /var/log/journal

These can then be reviewed after a crash via:

sudo journalctl

Watch out for kernel errors, which will appear in red colour in the journal output. Probably you can identify the culprit service, cron job or similar, that started or did a task before the crash, or indeed a voltage or temperature warning.

1
  • I'm accepting this because it's the closest to what I originally asked. However, these logs didn't contain anything illumination. I believe the issue was the power supply, since I have swapped that out for an old one, and the problem seems to have disappeared. Nov 29, 2021 at 2:32
0

I would suspect either the operating system or maybe the PC's BIOS to have detected some hazardous 'situation', most probably temperature but maybe also some voltage out of range.

It could also be a result of the battery on the motherboard failing.

  • I'd start measuring the battery voltage (take it out to do that. Most batteries are 3V type, and should measure above 3V when not inserted. When inserted it shouldn't drop voltage (but it's difficult to measure it).

  • While OFF, unplug the power cable to the hard disk (so the machine won't boot up). Power on, and enter the BIOS, check if the machine alone has the same problem. While waiting, check the hardware page of the BIOS (it should show voltages and temps. Voltages should be within 5% of their nominal values. Check if temp limits are reasonable, and that temp units are correct (Fahrenheit vs Celsius!)) Monitor temps for minutes and check if temps stabilize at a reasonable (low) value. If temp continues to rise maybe CPU heat sink is not correctly mounted?

  • Re-connect the power to the disk (while OFF). If the problem re-appears, maybe the extra load is making the CPU temp to rise. Enter the Debian hardware temp monitor program as quicky as possible and monitor temps and voltages for anomalies (both the 'set' limits and the actual values)

In any case, it's probable that the issue is power supply, or CPU temp related. Make a note of the ambient temp and check if the system works longer with cool ambient temp.

1
  • Thank you, this was very helpful advice. It appears that for my motherboard (Aorus X570) this is a common issue. However, it was not the source of my trouble. I replaced the CMOS battery the problem still occurred. Replacing the power supply with an older one, however, did fix the issue. Nov 29, 2021 at 2:33

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .