630-GPU服务器宕机，自动重启,日志记录:A fatal error was detected on a component at bus 128 device 3 function 0
故障原因：
造成机器宕机的原因是当多GPU高负载工作时， GPU 温度达到阈值（95度）触发了bus fatal error，导致重启宕机。
根本原因是IDRAC 温控进程异常，无法准确实时的反馈GPU实际工作温度，从而使GPU过热宕机；
Racadm直接调整风扇转速方式：
查看当前值：
[root@xxxxx ~]#racadm -r BMCIP -u xxx -p xxx get System.ThermalSettings.FanSpeedoffset
Security Alert: Certificate is invalid - self signed certificate
Continuing execution. Use -S option for racadm to stop execution on certificate-related errors.
[Key=System.Embedded.1#ThermalSettings.1]
FanSpeedOffset=Off设置风扇转速值为3：【0 low fan speed、1 medium fan speed、2 high fan speed、3 max fan speed】
[root@xxxxx ~]# racadm -r BMCIP -u xxx -p xxx set System.ThermalSettings.FanSpeedoffset 3
Security Alert: Certificate is invalid - self signed certificate
Continuing execution. Use -S option for racadm to stop execution on certificate-related errors.
[Key=System.Embedded.1#ThermalSettings.1]
Object value modified successfully设置完成后再次查看：
[root@xxxxx ~]#racadm -r BMCIP -u xxx -p xxx get System.ThermalSettings.FanSpeedoffset
Security Alert: Certificate is invalid - self signed certificate
Continuing execution. Use -S option for racadm to stop execution on certificate-related errors.
[Key=System.Embedded.1#ThermalSettings.1]
FanSpeedOffset=Max Fan Speed通过调整风扇转速，服务器运行正常。
文章来源：https://www.cnaaa.net，转载请注明出处：https://www.cnaaa.net/archives/8190
