Finding reason for system restart?

[TreyH]

If this is an FAQ, pardon me—I couldn’t figure out the search terms to unearth it. My Unraid host rebooted 2 days ago (on July 26), and I was trying to figure out why. Prior to that, my uptime had been about 3 months.

 

My husband said he installed Windows Updates in the Win10 guest VM, requiring reboots at about that time (not sure exactly when, though—it could have been before or during the host reboot). He says he didn’t touch a reset button or power switch/cable.

 

I’ve never seen action on the Win10 VM cause a reboot of Unraid—that VM is “his PC” so I barely ever touch it—so I’m perplexed. My vfio-pci log, dmesg, and syslog all start at the same time uptime gives me, and prior to that in /var/log I have nothing between mid-April and Jul 26. I do have guest support installed such that if I command Unraid to reboot, it suspends the Win10 VM gracefully and restarts it when Unraid comes back, and that has worked flawlessly in the past.

 

So all this sounds to me like it’s pointing to an unexpected powercycle—except

1. We were both home then and would normally have noticed the reboot chirp, and
2. Windows says its VM’s uptime is the same minute as Unraid’s, and Windows’ diags says it was a normal reboot, not a powercycle crash.

 

So… very, very odd.

 

Can anyone suggest next diagnosis steps?

Edited July 28, 2022 by TreyH
the word diagnostics gets autolinked to the wrong document here

[trurl]

attach diagnostics to your NEXT post in this thread

[theruck]

similar question here. i just heard that my unraid server rebooted and i cannot find a reason while i know it was not me rebooting it or me having any automatic action that would do so

 

server-diagnostics-20231215-1149.zip

[JorgeB]

Server rebooting on its own, versus crashing or hanging is usually a hardware or power problem, you can enable the syslog server and post that after the next reboot, but if it's hardware likely there won't be anything there.

[trurl]

Have you done memtest?

[xokia]

On 12/15/2023 at 3:04 AM, JorgeB said:

Server rebooting on its own, versus crashing or hanging is usually a hardware or power problem, you can enable the syslog server and post that after the next reboot, but if it's hardware likely there won't be anything there.

my system updated to 6.12.6 yesterday and now randomly crashed and rebooted today. Was up for 120 days prior to that which was a manual reset.

 

I know you folks reset the log files on reboot. Instead of syslog server tracing to the same USB as the OS (this can be problematic if during the crash something gets corrupted, your usb can become unbootable) would be really useful if we had an option to log to a separate USB device. I would just leave the USB stick in and let it write and rewrite that USB. If it went bad from to many writes just swap a new one in.

Edited January 2 by xokia

[trurl]

32 minutes ago, xokia said:

you folks reset the log files on reboot. Instead of syslog server tracing to the same USB as the OS (this can be problematic if during the crash something gets corrupted, your usb can become unbootable) would be really useful if we had an option to log to

The log files are in RAM like the rest of the OS.

 

There are multiple options for syslog server, read the link.

[xokia]

2 hours ago, trurl said:

The log files are in RAM like the rest of the OS.

 

There are multiple options for syslog server, read the link.

aware of the different ways to log. To log locally you have to do "trickery" and cause unnecessary network thrash. To mirror to flash you run into the problem I just mentioned. The log can corrupt your OS.

 

Just offering a suggestion to make logging more user friendly and useful. Log to flash should have an option to log to a secondary flash device so corruption of the OS USB does not occur. Logging locally should not require "trickery" it should be a supported feature JMO. 

Edited January 2 by xokia