Hello

 

I have an Orangepi One which fails after running for a certain time. In the failed "state" pinging it still works, but SSH connections to the Orangepi or from the Orangepi to the Server via Autossh won't work any more. A PHP application running on it is still working. It looks like the issue is that it winds back the time to something ridiculous. Hard reset with pulling Power fixes this for again a certain time till it happens again.

 

I wrote a script which runs every Minute via Cronjob and logs the current time to a file. Before the rewind already something weird is happening. The cronjob is executed twice on 04.40. I have no idea what is happening there and if it is related to the rewinding.

 

Fri 30 Oct 2020 04:34:01 AM CET
Fri 30 Oct 2020 04:35:01 AM CET
Fri 30 Oct 2020 04:36:01 AM CET
Fri 30 Oct 2020 04:37:01 AM CET
Fri 30 Oct 2020 04:38:01 AM CET
Fri 30 Oct 2020 04:40:02 AM CET
Fri 30 Oct 2020 04:40:02 AM CET
Fri 30 Oct 2020 04:41:01 AM CET
Tue 13 Nov 1979 05:23:13 PM CET
Tue 13 Nov 1979 05:24:01 PM CET
Tue 13 Nov 1979 05:25:01 PM CET
Tue 13 Nov 1979 05:26:01 PM CET
Tue 13 Nov 1979 05:27:01 PM CET
Tue 13 Nov 1979 05:28:02 PM CET
Tue 13 Nov 1979 05:29:01 PM CET
Tue 13 Nov 1979 05:30:01 PM CET

 

I solved this in a way that I fixed the cpu speed on one frequency! This problem never happend again after one year!

Sent from my MI 6 using Tapatalk

Odd. Never had this issue on any of my OPi1 boards. Or never noticed

I set up the same cronjob on one of my boards running legacy kernel and see if it is reproducible. Have my doubts though.

Ok, I have now fixed my cpu to 816 Mhz. Let's see if it now runs stable..

Ok, I have now fixed my cpu to 816 Mhz. Let's see if it now runs stable..

im holding mine on 1.37GHz

Sent from my MI 6 using Tapatalk

Does thermal throttling work? At night cronjobs start which might overheat the board.

Nonetheless, /etc/fake-hwclock.data should update by the hour, could you include hwclock -r and /etc/fake-hwclock.data?

 

Did now collect the date every minute for multiple days and could not detect any weirdness.

Hi,

   My orange pi PC and orange pi one used to crash like this.  It was fixed by running a more recent linux kernel, I recall it being stable in linux kernel 5.7 and above. 

 

From your pastelog, i see you are running an older version of armbian and using 5.4 linux kernel.  My crashing back to 197X year problem went away I think after upgrading to 5.7 linux kernel or above.  It used to hang and reboot into  year 197X every 3 to 17 days.  I can confirm fixed cpu speed made it stable but with the current mainline kernel there is no need for this anymore.

Really odd. I run my OPi One with PiHole on 5.4 and never had issues for a year now...

I only encountered the 197X issue on my allwinner h3 based devices.  I have

1X orange pi pc (h3)

2X orange pi one (h3)

3X banana pi M1 (a20)

1X orange pi PC 2 (h5)

 

The h3 devices were crashing to 197X and I used to reboot them daily.

They are all working fine now with 5.8 kernels and stable.

Little Update from my side. Had the probably same issue today on a OrangePi PC. SSH wasn't working any more, Ping worked. After Hard Reset everything fine again. This Board is at another Geo Location with different Power Supply. I have now installed the logging Cronjob + Telnet to check if it is really this issue if it happens again.. (This Board had 5.4.20-sunxi #20.02.1 installed, will update now)