SG1100 Panic：CPU 未响应 MCE 广播

• 管理节点（SG1100）发生意外重新启动和暂时无法访问。

• BMC 日志 CPU 目录错误（CATERR）

 331  Mar/13/2026   01:37:41  [Information] [Host Res Warning] [OEM]  Host Partition Reset triggered 255 minutes - Asserted
 330  Mar/13/2026   01:36:37  [Critical]  [CATERR]  [Processor] IERR - Asserted
 329  Mar/13/2026   01:35:10  [Critical]  [CATERR]  [Processor] Machine Check Exception (MCERR) - Asserted

• storagegrid_crash_dmesg.log 表示 kernel 由于 CPUs not responding to MCE broadcast 触发了恐慌

 [5048608.845286] watchdog: BUG: soft lockup - CPU#75 stuck for 78s! [prometheus-node:46612]
...
 [5048616.006133] mce: CPUs not responding to MCE broadcast (may include false positives): 10,58
 [5048616.006138] Kernel panic - not syncing: Timeout: Not all CPUs entered broadcast exception handler