Armbian 24.2 and Orange Pi One

[ER Samson]

Seems like the OS "malfunctions" after a few hours to a couple of days. The server is still responding to Pings, but services are all down, such as SSH.  Doing a power cycle would make the system available back online, until it becomes non-responsive again.  Moving back to 23.11.1 solves the stability issue.  To add, I'm using a couple of instances for primary and secondary PiHole DNS servers.

 

Probably similar issue on this thread -

 

I'm a bit of a noob, if someone can redirect me to pasting logs. Can those be found on /var/logs as I haven't explored much.  I have downgraded both my OPiOnes to 23.11.1 and Frozen the Armbian firmware updates, so I can't provide logs for now at least until I try to update one of the instances.

 

 

Edited March 24 by ER Samson

[Stephen Graf]

For a quick test I ran sbc-bench every 20 minutes overnight, about 11 hours. The system is still running well.

See the attached log.

sbc-bench v0.9.65 Xunlong Orange Pi.txt

 

You might want to connect to the serial port logged in as root to see if it is available after a system hang.  There may be some messages.

[ER Samson]

I may have found the culprit after monitoring htop. "vnstat -n" process eats up CPU cycles, sometimes reaching up to "6+" load-average (even if this OrangePi is quad-core equipped) 

 

This only happens intermittently and I can't seem to find the pattern.  Sharing my observations here:

• Only affects  Linux 6.6.16-current-sunxi and higher (such as edge), legacy firmware is not affected (stable for more than a week)
• Installing Pihole with cloudflared DoH package seems to trigger high CPU usage of vnstat
• Reboot fixes it (or a power cycle)
• Having PiHole with dnscrypt-proxy does not trigger the issue (so far, this test instance is already at the 16-hr mark, will keep on updating uptime)
  • Since it was stable overnight I now have two instances with this configuration, hoping for the best! 

Not sure if helpful, but this is all that I have for now. Thanks!