For some time now I'm getting random system crashes and I cannot pin down the cause. This happens usually in morning hours in highly irregular intervals: it happened today about 7am, and last time was about 3 weeks ago, before that it did not crash for 3 months but before that it was crashing every few days.

 

I'm attaching server diagnostics file and syslog from today (I'm storing syslog in share, so its not lost on a crash).

 

What can I do to find the cause of this?

beholder-diagnostics-20240430-0850.zip syslog.log

Apr 29 23:50:54 Beholder kernel: [Hardware Error]: Corrected error, no action required.
Apr 29 23:50:54 Beholder kernel: [Hardware Error]: CPU:0 (17:71:0) MC27_STATUS[-|CE|MiscV|-|-|-|SyndV|-|-|-]: 0x982000000002080b
Apr 29 23:50:54 Beholder kernel: [Hardware Error]: IPID: 0x0001002e00000500, Syndrome: 0x000000005a020001
Apr 29 23:50:54 Beholder kernel: [Hardware Error]: Power, Interrupts, etc. Ext. Error Code: 2, Link Error.
Apr 29 23:50:54 Beholder kernel: [Hardware Error]: cache level: L3/GEN, mem/io: IO, mem-tx: GEN, part-proc: SRC (no timeout)

 

This suggests a CPU problem, this time the error was corrected, but sometimes it may not be, and cause a crash.

On 4/30/2024 at 11:39 AM, JorgeB said:

Apr 29 23:50:54 Beholder kernel: [Hardware Error]: Corrected error, no action required.
Apr 29 23:50:54 Beholder kernel: [Hardware Error]: CPU:0 (17:71:0) MC27_STATUS[-|CE|MiscV|-|-|-|SyndV|-|-|-]: 0x982000000002080b
Apr 29 23:50:54 Beholder kernel: [Hardware Error]: IPID: 0x0001002e00000500, Syndrome: 0x000000005a020001
Apr 29 23:50:54 Beholder kernel: [Hardware Error]: Power, Interrupts, etc. Ext. Error Code: 2, Link Error.
Apr 29 23:50:54 Beholder kernel: [Hardware Error]: cache level: L3/GEN, mem/io: IO, mem-tx: GEN, part-proc: SRC (no timeout)

 

This suggests a CPU problem, this time the error was corrected, but sometimes it may not be, and cause a crash.

So your recommendation is to replace my CPU?

If you can try a different one first to confirm, I can't say for sure that the CPU is the problem, but it would be my main suspect.