HP proliant（2 CPU）上的 Debian 6，配备 raid（2*1.5T RAID1 + 2*2T RAID1 加入 RAID0 以构成 3.5T），主要运行 nfs 和 imapd（加上用于 Windows 共享的 samba 和用于预览网页的本地 www）；使用本地 ubuntu 桌面客户端安装 $HOME，笔记本电脑通过 nfs/smb 访问 imap 和奇数文件（例如视频）；盒子通过家庭路由器/交换机连接到 100baseT 或 wifi
uname -a
Linux prole 2.6.32-5-686 #1 SMP Wed Jan 11 12:29:30 UTC 2012 i686 GNU/Linux
设置已经工作了几个月，但容易间歇性地变得非常缓慢（从服务器安装桌面 $HOME 或笔记本电脑播放视频的用户体验）并且现在一直很糟糕，我不得不深入研究它以试图找出问题所在（！）
服务器在低负载下似乎没问题，例如（笔记本电脑）客户端（本地磁盘上的 $HOME）连接到服务器的 imapd 和 nfs 安装 RAID 来访问 1 个文件：顶部显示负载约为 0.1 或更低，0 等待
但是当（桌面）客户端挂载 $HOME 并启动用户 KDE 会话（所有访问服务器）时顶部例如
top - 13:41:17 up 3:43, 3 users, load average: 9.29, 9.55, 8.27
Tasks: 158 total, 1 running, 157 sleeping, 0 stopped, 0 zombie
Cpu(s): 0.4%us, 0.4%sy, 0.0%ni, 49.0%id, 49.7%wa, 0.0%hi, 0.5%si, 0.0%st
Mem: 903856k total, 851784k used, 52072k free, 171152k buffers
Swap: 0k total, 0k used, 0k free, 476896k cached
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
3935 root 20 0 2456 1088 784 R 2 0.1 0:00.02 top
1 root 20 0 2028 680 584 S 0 0.1 0:01.14 init
2 root 20 0 0 0 0 S 0 0.0 0:00.00 kthreadd
3 root RT 0 0 0 0 S 0 0.0 0:00.00 migration/0
4 root 20 0 0 0 0 S 0 0.0 0:00.12 ksoftirqd/0
5 root RT 0 0 0 0 S 0 0.0 0:00.00 watchdog/0
6 root RT 0 0 0 0 S 0 0.0 0:00.00 migration/1
7 root 20 0 0 0 0 S 0 0.0 0:00.16 ksoftirqd/1
8 root RT 0 0 0 0 S 0 0.0 0:00.00 watchdog/1
9 root 20 0 0 0 0 S 0 0.0 0:00.42 events/0
10 root 20 0 0 0 0 S 0 0.0 0:02.26 events/1
11 root 20 0 0 0 0 S 0 0.0 0:00.00 cpuset
12 root 20 0 0 0 0 S 0 0.0 0:00.00 khelper
13 root 20 0 0 0 0 S 0 0.0 0:00.00 netns
14 root 20 0 0 0 0 S 0 0.0 0:00.00 async/mgr
15 root 20 0 0 0 0 S 0 0.0 0:00.00 pm
16 root 20 0 0 0 0 S 0 0.0 0:00.02 sync_supers
17 root 20 0 0 0 0 S 0 0.0 0:00.02 bdi-default
18 root 20 0 0 0 0 S 0 0.0 0:00.00 kintegrityd/0
19 root 20 0 0 0 0 S 0 0.0 0:00.00 kintegrityd/1
20 root 20 0 0 0 0 S 0 0.0 0:00.02 kblockd/0
21 root 20 0 0 0 0 S 0 0.0 0:00.08 kblockd/1
22 root 20 0 0 0 0 S 0 0.0 0:00.00 kacpid
23 root 20 0 0 0 0 S 0 0.0 0:00.00 kacpi_notify
24 root 20 0 0 0 0 S 0 0.0 0:00.00 kacpi_hotplug
25 root 20 0 0 0 0 S 0 0.0 0:00.00 kseriod
28 root 20 0 0 0 0 S 0 0.0 0:04.19 kondemand/0
29 root 20 0 0 0 0 S 0 0.0 0:02.93 kondemand/1
30 root 20 0 0 0 0 S 0 0.0 0:00.00 khungtaskd
31 root 20 0 0 0 0 S 0 0.0 0:00.18 kswapd0
32 root 25 5 0 0 0 S 0 0.0 0:00.00 ksmd
33 root 20 0 0 0 0 S 0 0.0 0:00.00 aio/0
34 root 20 0 0 0 0 S 0 0.0 0:00.00 aio/1
35 root 20 0 0 0 0 S 0 0.0 0:00.00 crypto/0
36 root 20 0 0 0 0 S 0 0.0 0:00.00 crypto/1
203 root 20 0 0 0 0 S 0 0.0 0:00.00 ksuspend_usbd
204 root 20 0 0 0 0 S 0 0.0 0:00.00 khubd
205 root 20 0 0 0 0 S 0 0.0 0:00.00 ata/0
206 root 20 0 0 0 0 S 0 0.0 0:00.00 ata/1
207 root 20 0 0 0 0 S 0 0.0 0:00.14 ata_aux
208 root 20 0 0 0 0 S 0 0.0 0:00.01 scsi_eh_0
消息表明存在磁盘问题：
.............. (previous episode)
[13276.966004] raid1:md0: read error corrected (8 sectors at 489900360 on sdc7)
[13276.966043] raid1: sdb7: redirecting sector 489898312 to another mirror
[13279.569186] ata4.00: exception Emask 0x0 SAct 0x1 SErr 0x0 action 0x0
[13279.569211] ata4.00: irq_stat 0x40000008
[13279.569230] ata4.00: failed command: READ FPDMA QUEUED
[13279.569257] ata4.00: cmd 60/08:00:00:6a:05/00:00:23:00:00/40 tag 0 ncq 4096 in
[13279.569262] res 41/40:00:05:6a:05/00:00:23:00:00/40 Emask 0x409 (media error) <F>
[13279.569306] ata4.00: status: { DRDY ERR }
[13279.569321] ata4.00: error: { UNC }
[13279.575362] ata4.00: configured for UDMA/133
[13279.575388] ata4: EH complete
[13283.169224] ata4.00: exception Emask 0x0 SAct 0x1 SErr 0x0 action 0x0
[13283.169246] ata4.00: irq_stat 0x40000008
[13283.169263] ata4.00: failed command: READ FPDMA QUEUED
[13283.169289] ata4.00: cmd 60/08:00:00:6a:05/00:00:23:00:00/40 tag 0 ncq 4096 in
[13283.169294] res 41/40:00:07:6a:05/00:00:23:00:00/40 Emask 0x409 (media error) <F>
[13283.169331] ata4.00: status: { DRDY ERR }
[13283.169345] ata4.00: error: { UNC }
[13283.176071] ata4.00: configured for UDMA/133
[13283.176104] ata4: EH complete
[13286.224814] ata4.00: exception Emask 0x0 SAct 0x1 SErr 0x0 action 0x0
[13286.224837] ata4.00: irq_stat 0x40000008
[13286.224853] ata4.00: failed command: READ FPDMA QUEUED
[13286.224879] ata4.00: cmd 60/08:00:00:6a:05/00:00:23:00:00/40 tag 0 ncq 4096 in
[13286.224884] res 41/40:00:06:6a:05/00:00:23:00:00/40 Emask 0x409 (media error) <F>
[13286.224922] ata4.00: status: { DRDY ERR }
[13286.224935] ata4.00: error: { UNC }
[13286.231277] ata4.00: configured for UDMA/133
[13286.231303] ata4: EH complete
[13288.802623] ata4.00: exception Emask 0x0 SAct 0x1 SErr 0x0 action 0x0
[13288.802646] ata4.00: irq_stat 0x40000008
[13288.802662] ata4.00: failed command: READ FPDMA QUEUED
[13288.802688] ata4.00: cmd 60/08:00:00:6a:05/00:00:23:00:00/40 tag 0 ncq 4096 in
[13288.802693] res 41/40:00:05:6a:05/00:00:23:00:00/40 Emask 0x409 (media error) <F>
[13288.802731] ata4.00: status: { DRDY ERR }
[13288.802745] ata4.00: error: { UNC }
[13288.808901] ata4.00: configured for UDMA/133
[13288.808927] ata4: EH complete
[13291.380430] ata4.00: exception Emask 0x0 SAct 0x1 SErr 0x0 action 0x0
[13291.380453] ata4.00: irq_stat 0x40000008
[13291.380470] ata4.00: failed command: READ FPDMA QUEUED
[13291.380496] ata4.00: cmd 60/08:00:00:6a:05/00:00:23:00:00/40 tag 0 ncq 4096 in
[13291.380501] res 41/40:00:05:6a:05/00:00:23:00:00/40 Emask 0x409 (media error) <F>
[13291.380577] ata4.00: status: { DRDY ERR }
[13291.380594] ata4.00: error: { UNC }
[13291.386517] ata4.00: configured for UDMA/133
[13291.386543] ata4: EH complete
[13294.347147] ata4.00: exception Emask 0x0 SAct 0x1 SErr 0x0 action 0x0
[13294.347169] ata4.00: irq_stat 0x40000008
[13294.347186] ata4.00: failed command: READ FPDMA QUEUED
[13294.347211] ata4.00: cmd 60/08:00:00:6a:05/00:00:23:00:00/40 tag 0 ncq 4096 in
[13294.347217] res 41/40:00:06:6a:05/00:00:23:00:00/40 Emask 0x409 (media error) <F>
[13294.347254] ata4.00: status: { DRDY ERR }
[13294.347268] ata4.00: error: { UNC }
[13294.353556] ata4.00: configured for UDMA/133
[13294.353583] sd 3:0:0:0: [sdc] Unhandled sense code
[13294.353590] sd 3:0:0:0: [sdc] Result: hostbyte=DID_OK driverbyte=DRIVER_SENSE
[13294.353599] sd 3:0:0:0: [sdc] Sense Key : Medium Error [current] [descriptor]
[13294.353610] Descriptor sense data with sense descriptors (in hex):
[13294.353616] 72 03 11 04 00 00 00 0c 00 0a 80 00 00 00 00 00
[13294.353635] 23 05 6a 06
[13294.353644] sd 3:0:0:0: [sdc] Add. Sense: Unrecovered read error - auto reallocate failed
[13294.353657] sd 3:0:0:0: [sdc] CDB: Read(10): 28 00 23 05 6a 00 00 00 08 00
[13294.353675] end_request: I/O error, dev sdc, sector 587557382
[13294.353726] ata4: EH complete
[13294.366953] raid1:md0: read error corrected (8 sectors at 489900544 on sdc7)
[13294.366992] raid1: sdc7: redirecting sector 489898496 to another mirror
而且它们发生得相当频繁，我猜这很可能会导致性能问题(?)
# dmesg | grep 镜像
[12433.561822] raid1: sdc7: redirecting sector 489900464 to another mirror
[12449.428933] raid1: sdb7: redirecting sector 489900504 to another mirror
[12464.807016] raid1: sdb7: redirecting sector 489900512 to another mirror
[12480.196222] raid1: sdb7: redirecting sector 489900520 to another mirror
[12495.585413] raid1: sdb7: redirecting sector 489900528 to another mirror
[12510.974424] raid1: sdb7: redirecting sector 489900536 to another mirror
[12526.374933] raid1: sdb7: redirecting sector 489900544 to another mirror
[12542.619938] raid1: sdc7: redirecting sector 489900608 to another mirror
[12559.431328] raid1: sdc7: redirecting sector 489900616 to another mirror
[12576.553866] raid1: sdc7: redirecting sector 489900624 to another mirror
[12592.065265] raid1: sdc7: redirecting sector 489900632 to another mirror
[12607.621121] raid1: sdc7: redirecting sector 489900640 to another mirror
[12623.165856] raid1: sdc7: redirecting sector 489900648 to another mirror
[12638.699474] raid1: sdc7: redirecting sector 489900656 to another mirror
[12655.610881] raid1: sdc7: redirecting sector 489900664 to another mirror
[12672.255617] raid1: sdc7: redirecting sector 489900672 to another mirror
[12672.288746] raid1: sdc7: redirecting sector 489900680 to another mirror
[12672.332376] raid1: sdc7: redirecting sector 489900688 to another mirror
[12672.362935] raid1: sdc7: redirecting sector 489900696 to another mirror
[12674.201177] raid1: sdc7: redirecting sector 489900704 to another mirror
[12698.045050] raid1: sdc7: redirecting sector 489900712 to another mirror
[12698.089309] raid1: sdc7: redirecting sector 489900720 to another mirror
[12698.111999] raid1: sdc7: redirecting sector 489900728 to another mirror
[12698.134006] raid1: sdc7: redirecting sector 489900736 to another mirror
[12719.034376] raid1: sdc7: redirecting sector 489900744 to another mirror
[12734.545775] raid1: sdc7: redirecting sector 489900752 to another mirror
[12734.590014] raid1: sdc7: redirecting sector 489900760 to another mirror
[12734.624050] raid1: sdc7: redirecting sector 489900768 to another mirror
[12734.647308] raid1: sdc7: redirecting sector 489900776 to another mirror
[12734.664657] raid1: sdc7: redirecting sector 489900784 to another mirror
[12734.710642] raid1: sdc7: redirecting sector 489900792 to another mirror
[12734.721919] raid1: sdc7: redirecting sector 489900800 to another mirror
[12734.744732] raid1: sdc7: redirecting sector 489900808 to another mirror
[12734.779330] raid1: sdc7: redirecting sector 489900816 to another mirror
[12782.604564] raid1: sdb7: redirecting sector 1242934216 to another mirror
[12798.264153] raid1: sdc7: redirecting sector 1242935080 to another mirror
[13245.832193] raid1: sdb7: redirecting sector 489898296 to another mirror
[13261.376929] raid1: sdb7: redirecting sector 489898304 to another mirror
[13276.966043] raid1: sdb7: redirecting sector 489898312 to another mirror
[13294.366992] raid1: sdc7: redirecting sector 489898496 to another mirror
尽管阵列仍在所有磁盘上运行 - 但它们尚未放弃任何磁盘：
# 猫 /proc/mdstat
Personalities : [raid1] [raid0]
md10 : active raid0 md0[0] md1[1]
3368770048 blocks super 1.2 512k chunks
md1 : active raid1 sde2[2] sdd2[1]
1464087824 blocks super 1.2 [2/2] [UU]
md0 : active raid1 sdb7[0] sdc7[2]
1904684920 blocks super 1.2 [2/2] [UU]
unused devices: <none>
所以我认为我知道问题是什么，但是我并不是一个 Linux 系统管理员专家，如果能提供一些线索来检查我的诊断以及我需要做什么，我将非常感激：
- 显然我需要为 sdc 寻找另一个驱动器。（我猜如果价格合适，我可以买一个更大的驱动器：我想有一天我需要增加阵列的大小，这样就不需要再用一个更大的驱动器来替换了）
- 然后使用 mdadm 使现有的 sdc 失效，移除它并安装新的驱动器
- 使用 fdisk 检查新驱动器，使其具有与旧驱动器相同的阵列分区大小
- 使用 mdadm 将新驱动器添加到阵列中
听起来还好吗？
答案1
通常，当出现您看到的磁盘错误时，磁盘会暂停一会儿，尝试自行纠正错误，而 Linux RAID 会容忍某些磁盘等待，直到将其标记为错误为止。磁盘暂停可能是导致速度变慢的原因（尤其是在您看到的错误率下）。
您更换驱动器的计划是正确的。但是，我不建议购买更大的驱动器，因为这样可以将一部分分区用于替换 RAID，然后将另一部分分区用于以后的其他用途。更明智的做法是更接近原始磁盘（大小和速度）以保持阵列一致。也就是说，理论上，您可以扩大驱动器，将确切的替换大小分区用于阵列替换，然后将另一个分区用于其他内容（甚至是阵列的另一个成员）。
可能有助于调试的附注：我喜欢用来替代 top 的工具叫做 atop (http://www.atoptool.nl/），它可以为您提供更直观的视图，让您了解每个磁盘和使用磁盘 I/O 的进程以及瓶颈所在（您可能会注意到等待 I/O 是针对出现问题的特定磁盘的）。