troubleshooting or "the king is dead long live the king!"

My jetson has got to brick after two times suddenly getting off without an explicit reason.
[red led lights, but thats it] It turned on a couple of times for few minutes and got to a state when it is not powering on anymore, as it seems to me.
Hm, seems it got on again after disconnecting ac/dc for few minutes and plugging it into the board.
What kind of errors should I look within the dmsg or log files to distinguish the reason?
Though, it is getting suddenly powering down again.
Warranty got expired couple of month ago.
Well, it appears that first I shall approach re flashing it…
seems to start up at once,
seems working
seems resolved
update: it worked for 3 hours and got powered off again :(
are there any ways for troubleshooting? Any way to distinguish the cause to investigate the issue of resolution of the cause in case the latter got distinguished?
Or I shall just consider “the king is dead long live the king” and get a new model as soon as it got released? When xavier will be available for ordering?

Once it powers off can you clone the rootfs? You might want the clone, but a reason for attempting this (beyond seeing data) is to see if the lower level functions to read eMMC still work.

Can you get it to run with serial console connected and log the serial console until it dies?

FYI, when you go to restart a Jetson, and if U-Boot succeeds in loading, then this next boot after a failure could provide valuable logs as well since it might show some “reason” information as to what caused the reboot.

Just speculating, but one possible cause (even if the Jetson hardware is good) is if there is some brief power spike which is too fast for the human eye to see (marginal power supplies might do this and leave something corrupted).

After flashing with Jetpack 3.3:

 10:06:35 up  2:02,  2 users,  load average: 2.31, 1.73, 1.72

I did not enable “” and “nvpm 0” though

Does that mean the system is stable after flashing with 3.3? It sounds like the issue is resolved, but I’m not sure if 10 hours was the failure time or just shows the system kept running successfully.

Unfortunately not, it worked for about 20 hours and got getting powering off on itself [may be it got to work longer because I did not use nvpm & jetsonclock].
However, I shall test some another equivalent power supply, in my opinion. To exclude the power-supply cause.

Do you see any traces of the shutdown in /var/log/syslog or /var/log/kern.log ?
If it is a power supply problem, you might see nothing logged, but if it is a thermal or software problem, you may get some clues there.