我一直在凌晨发生系统崩溃。00:00 至 08:00 之间。
查看日志,看起来每个核心都慢了下来,最终崩溃,直到整个系统冻结。从第一个错误到系统崩溃大约需要 25 分钟。这些错误让我想到了内存问题,但我不确定这是否只是因为其他一些故障。
该系统是具有双 E5-2630 V2 和 16X 8GB DDR3 的 Super Micro X9DRW-IF
操作系统是 Proxmox 最新的
核心:
Linux pve1 4.15.18-10-pve #1 SMP PVE 4.15.18-32 (Sat, 19 Jan 2019)
PERF 中断开始花费越来越长的时间,并且内存错误开始发生。在系统在 20 分钟到 1 小时内冻结之前,大约会出现 20 个此类错误。据我了解,PERF 只是 CPU 节流。它会限制到尽可能低的速度,此时系统会爬行。
Apr 28 07:36:05 pve1 kernel: [36497.018818] perf: interrupt took too long (6737393 > 4247631), lowering kernel.perf_event_max_sample_rate to 250
Apr 28 07:36:05 pve1 kernel: [36497.018914] {1}[Hardware Error]: Hardware error from APEI Generic Hardware Error Source: 1
Apr 28 07:36:05 pve1 kernel: [36497.018926] {1}[Hardware Error]: It has been corrected by h/w and requires no further action
Apr 28 07:36:05 pve1 kernel: [36497.019012] {1}[Hardware Error]: event severity: corrected
Apr 28 07:36:05 pve1 kernel: [36497.019112] {1}[Hardware Error]: Error 0, type: corrected
Apr 28 07:36:05 pve1 kernel: [36497.019115] {1}[Hardware Error]: fru_text: CorrectedErr
Apr 28 07:36:05 pve1 kernel: [36497.019119] {1}[Hardware Error]: section_type: memory error
Apr 28 07:36:05 pve1 kernel: [36497.019125] {1}[Hardware Error]: node: 1 device: 0
Apr 28 07:36:05 pve1 kernel: [36497.019128] {1}[Hardware Error]: error_type: 2, single-bit ECC
Apr 28 07:36:05 pve1 kernel: [36497.019297] ghes_edac: Internal error: Can't find EDAC structure
Apr 28 07:36:06 pve1 pve-firewall[2311]: firewall update time (13.994 seconds)
Apr 28 07:36:10 pve1 kernel: [36502.054892] INFO: NMI handler (perf_event_nmi_handler) took too long to run: 451.489 msecs
Apr 28 07:36:17 pve1 pve-firewall[2311]: firewall update time (9.985 seconds)
Apr 28 07:36:20 pve1 pvestatd[2315]: got timeout
Apr 28 07:36:26 pve1 pvestatd[2315]: status update time (33.041 seconds)
Apr 28 07:36:28 pve1 pve-firewall[2311]: firewall update time (11.073 seconds)
Apr 28 07:36:50 pve1 kernel: [36542.038771] INFO: NMI handler (perf_event_nmi_handler) took too long to run: 451.686 msecs
Apr 28 07:36:56 pve1 pve-firewall[2311]: firewall update time (27.943 seconds)
Apr 28 07:36:56 pve1 pvestatd[2315]: status update time (30.979 seconds)
Apr 28 07:37:03 pve1 pve-firewall[2311]: firewall update time (6.031 seconds)
在这一点上,我只想了解实际发生的情况。
我的托管主机拉出服务器并重新安装了所有 DIMM。
这已经3天没有发生了。似乎由于缺乏可靠的连接而导致连接变慢,越来越多的数据变得损坏,直到系统崩溃试图跳过错误的地址。松动的连接也可能已经开始降低内存电压,从而导致波动。电压保持在范围内,但随着这种情况开始发生,电压会变得不稳定。