论坛徽章:: 0

电梯直达

1楼 [收藏(0)] [报告]

发表于 2007-05-13 04:51 |只看该作者 |倒序浏览

2x3320（JBOD）x9块硬盘
SDS做raid0+1,如下：
# metastat -s dgsmp

dgsmp/d100: Mirror
Submirror 0: dgsmp/d101
   State: Needs maintenance
Submirror 1: dgsmp/d102
   State: Okay
Pass: 1
Read option: roundrobin (default)
Write option: parallel (default)
Size: 573356544 blocks

dgsmp/d101: Submirror of dgsmp/d100
State: Needs maintenance
Invoke: metareplace dgsmp/d100 d23s0 <new device>
Size: 573356544 blocks
Stripe 0: (interlace: 32 blocks)
      Device  Start Block  Dbase State       Hot Spare
      d20s0       0    No Maintenance
      d21s0       0    No Maintenance
      d22s0       0    No Maintenance
      d23s0       0    No Maintenance
dgsmp/d102: Submirror of dgsmp/d100
State: Okay
Size: 573356544 blocks
Stripe 0: (interlace: 32 blocks)
      Device  Start Block  Dbase State       Hot Spare
      d14s0       0    No Okay
      d15s0       0    No Okay
      d16s0       0    No Okay
      d17s0       0    No Okay

dgsmp/d120: Mirror
Submirror 0: dgsmp/d121
   State: Needs maintenance
Submirror 1: dgsmp/d122
   State: Needs maintenance
Pass: 1
Read option: roundrobin (default)
Write option: parallel (default)
Size: 716695680 blocks

dgsmp/d121: Submirror of dgsmp/d120
State: Needs maintenance
Invoke: metareplace dgsmp/d120 d19s0 <new device>
Size: 716695680 blocks
Stripe 0: (interlace: 32 blocks)
      Device  Start Block  Dbase State       Hot Spare
      d6s0          0    No Maintenance
      d7s0          0    No Maintenance
      d8s0          0    No Maintenance
      d18s0       0    No Maintenance
      d19s0       0    No Maintenance

dgsmp/d122: Submirror of dgsmp/d120
State: Needs maintenance
Invoke: after replacing "Maintenance" components:
            metareplace dgsmp/d120 d13s0 <new device>
Size: 716695680 blocks
Stripe 0: (interlace: 32 blocks)
      Device  Start Block  Dbase State       Hot Spare
      d9s0          0    No Okay
      d10s0       0    No Okay
      d11s0       0    No Okay
      d12s0       0    No Okay
      d13s0       0    No Last Erred

dgsmp/d100 -m dgsmp/d101 dgsmp/d102 1
dgsmp/d101 1 4 d20s0 d21s0 d22s0 d23s0 -i 32b
dgsmp/d102 1 4 d14s0 d15s0 d16s0 d17s0 -i 32b
dgsmp/d120 -m dgsmp/d121 dgsmp/d122 1
dgsmp/d121 1 5 d6s0 d7s0 d8s0 d18s0 d19s0 -i 32b
dgsmp/d122 1 5 d9s0 d10s0 d11s0 d12s0 d13s0 -i 32b

messages:
May  7 02:05:46 sdp03 scsi: [ID 107833 kern.warning] WARNING: /pci@1c,600000/scsi@1/sd@9,0 (sd99):
May  7 02:05:46 sdp03 Error for Command: read(10)             Error Level: Retryable
May  7 02:05:46 sdp03 scsi: [ID 107833 kern.notice] Requested Block: 12466784                Error Block: 12466803
May  7 02:05:46 sdp03 scsi: [ID 107833 kern.notice] Vendor: SEAGATE                         Serial Number: 055033E9RM
May  7 02:05:46 sdp03 scsi: [ID 107833 kern.notice] Sense Key: Media Error
May  7 02:05:46 sdp03 scsi: [ID 107833 kern.notice] ASC: 0x11 (unrecovered read error), ASCQ: 0x0, FRU: 0xf

iostat -E:
sd99    Soft Errors: 0 Hard Errors: 730 Transport Errors: 91
Vendor: SEAGATE  Product: ST373207LSUN72G  Revision: 045A Serial No: 5033E9RM
Size: 73.40GB <73400057856 bytes>
Media Error: 636 Device Not Ready: 0 No Device: 94 Recoverable: 0
Illegal Request: 0 Predictive Failure Analysis: 0

diskinfo:
c7t9d0 SEAGATE ST373207LSUN72G 045A 055033E9RM

scdidadm-L:
13    sdp03:/dev/rdsk/c7t9d0       /dev/did/rdsk/d13
13    sdp04:/dev/rdsk/c7t9d0       /dev/did/rdsk/d13

分析结论：
SE3320 一台所有硬盘Needs maintenance
Se3320 另一台有一块故障硬盘c7t9(/dev/did/rdsk/d13 )状态Last Erred
处理思路：
将Needs maintenance 的Se3320与另一台okey的分别同步，恢复成okey；
将Last Erred 的故障硬盘认为是正常的，尝试与d19s0 同步，估计同步会报I/O读写错误自动中断；
如果同步失败，将该盘换掉，与d19s0再同步成okey，再将数据倒回来。
预计d13s0数据会丢失，进行raid重建，并根据事先备份恢复数据
根据"/node@1/pci@1c,600000/scsi@1/sd@9,0" 99 "sd" 以及查询scsi卡端口物理路径，scsi cable连接，能判断出哪台3320有坏盘。
问题：
1 处理思路是否可行？因为是raid0,更换一块硬盘会导致数据丢失，能否根据另一台3320 mirror保证数据不丢？
2 Se3320 (JBOD)打开面板后，看不到硬盘序列号，如何来判断故障硬盘是哪个？硬盘状态灯都是正常绿的。