遭遇小概率故障:FC交换机端口或SFP模块坏

昨晚重启一台Linux服务器发现接EMC CX存储挂载的盘没有mount起来，感觉奇怪，因为前一次reboot系统还看到mount起来，一转眼就没了，通过查看HBA卡信息，发现和CX的连接很不稳定，时断时续，具体表现在一个很短时间内(比如10秒钟)，有时看到CX控制器port的WWN，有时则看不到，如下：

[root@localhost ~]# cat /proc/scsi/lpfc/2
[root@localhost ~]# cat /proc/scsi/lpfc/2
[root@localhost ~]# cat /proc/scsi/lpfc/2
[root@localhost ~]# date
Thu Jul 29 01:41:07 CST 2010
[root@localhost ~]# cat /proc/scsi/lpfc/2
lpfc1t00 DID 140900 WWPN 50:06:01:69:39:a0:55:2b WWNN 50:06:01:60:b9:a0:55:2b
lpfc1t01 DID 140a00 WWPN 50:06:01:61:39:a0:55:2b WWNN 50:06:01:60:b9:a0:55:2b
[root@localhost ~]# cat /proc/scsi/lpfc/2
[root@localhost ~]# date
Thu Jul 29 01:41:15 CST 2010

到机房去检查服务器发现，明显该HBA卡那个橙色灯闪烁不正常，亮两下，灭两下，而其他正常HBA卡灯持续闪烁，然后我倾向交换机那一端不稳定，到brocade那头，插拔了光纤线，结果照旧，把SFP模块再拔下重插一次也没效果，只好更换一个端口了，结果接上去HBA卡一端检查，橙色灯闪烁正常，再去OS上检查，也正常了

[root@localhost ~]# date
Thu Jul 29 01:57:43 CST 2010
[root@localhost ~]# cat /proc/scsi/lpfc/2
lpfc1t00 DID 140900 WWPN 50:06:01:69:39:a0:44:2b WWNN 50:06:01:60:b9:a0:55:2b
lpfc1t01 DID 140a00 WWPN 50:06:01:61:39:a0:44:2b WWNN 50:06:01:60:b9:a0:55:2b
[root@localhost ~]# cat /proc/scsi/lpfc/2
lpfc1t00 DID 140900 WWPN 50:06:01:69:39:a0:44:2b WWNN 50:06:01:60:b9:a0:55:2b
lpfc1t01 DID 140a00 WWPN 50:06:01:61:39:a0:44:2b WWNN 50:06:01:60:b9:a0:55:2b
[root@localhost ~]# cat /proc/scsi/lpfc/2
lpfc1t00 DID 140900 WWPN 50:06:01:69:39:a0:44:2b WWNN 50:06:01:60:b9:a0:55:2b
lpfc1t01 DID 140a00 WWPN 50:06:01:61:39:a0:44:2b WWNN 50:06:01:60:b9:a0:55:2b
[root@localhost ~]# cat /proc/scsi/lpfc/2
lpfc1t00 DID 140900 WWPN 50:06:01:69:39:a0:44:2b WWNN 50:06:01:60:b9:a0:55:2b
lpfc1t01 DID 140a00 WWPN 50:06:01:61:39:a0:44:2b WWNN 50:06:01:60:b9:a0:55:2b
[root@localhost ~]# cat /proc/scsi/lpfc/2
lpfc1t00 DID 140900 WWPN 50:06:01:69:39:a0:44:2b WWNN 50:06:01:60:b9:a0:55:2b
lpfc1t01 DID 140a00 WWPN 50:06:01:61:39:a0:44:2b WWNN 50:06:01:60:b9:a0:55:2b
[root@localhost ~]# date
Thu Jul 29 01:57:51 CST 2010

这个故障最大可能就是SFP模块坏了，也有可能是交换机端口坏，但可能性较小，因交换机大多是多个端口在一块板上，一坏可能就不是一个端口了