0

I have a supermicro 1U running Ubuntu 16.04.6 LTS. The problem I am having is it reboots at what appears to be random intervals; it can go 2 days between reboots or it can go 2 weeks between reboots. It is a plain vanilla LAMP server with high frequency write requests to MySql. The average load to the machine is about 0.09, so it's certainly not under any heavy load. No crons are set up via crontab.

What I have looked at:

  • Automatic updates are disabled. Besides, I am not aware of an OS that updates every 3 or 4 days so I am ruling this one out.
  • None of the log files (kern.log, syslog, as well as mysql and apache logs) show anything immediately prior to reboot. It's as if someone pulls the power on the server and then plugs it back in. There are NO errors logged right before the reboot.
  • The mysql log show this after reboot: 2021-08-13T05:31:51.164559Z 0 [Note] InnoDB: Database was not shutdown normally! 2021-08-13T05:31:51.164564Z 0 [Note] InnoDB: Starting crash recovery.. So we do know it wasn't a graceful shutdown.
  • I have swapped the RAM on the machine about a week ago to brand new Samsung sticks tested my SuperMicro with this specific motherboard, but the problem persists.
  • IPMI logs do not have anything pointing to any problems. In fact nothing is logged around the time of the reboot.

I am really at a loss here. Could someone please suggest anything else I can look at?

codemonkey
  • 171
  • 6
  • Go into the IPMI and check the health event log. – Michael Hampton Aug 13 '21 at 18:56
  • @MichaelHampton IPMI heath log's last entry is from a week ago when I swapped RAM. Nothing after that. Another reboot just happened this morning. – codemonkey Aug 13 '21 at 19:01
  • Interesting. And did you also check the sensors? – Michael Hampton Aug 13 '21 at 19:33
  • Sensors are all green. No logs of them being out of order. – codemonkey Aug 13 '21 at 19:47
  • Well I would suggest you first check the sensor thresholds to make sure they make sense, then set up remote syslog logging, or email alerts, or both. Maybe some clue will come in the next time it happens. I was thinking you have a hardware fault, but something should have been in the health event log when it rebooted. It's possible you instead have a software fault, and since your distro is past EOL, you might not be able to do anything about that except upgrade to a supported distro. – Michael Hampton Aug 13 '21 at 19:54

0 Answers0