我的盒子里有 6 个 HDD，2x256Gb ATA OCZ-AGILITY4 SSD。
- 每个分区都划分为一个 48Gb 的块和一个 208Gb 的块。
- 48Gb 分区是 RAID0 条带，并用作交换空间 (md0)。
- 208Gb 分区是 RAID1 条带，并用作 / 文件系统 (md1)。
还有 4x3.0Tb ATA ST3000DM001-9YN166 驱动器，每个驱动器都有一个分区，这 4 个分区是 RAID5 条带 (md128)。所有驱动器都是 ext4 格式，运行 ubuntu 12.04 服务器。sda 和 sdb 是 SSD，而 sdc、sdd、sde 和 sdf 是 HDD。
随机地（据我所知）md128 变为只读。事件的系统日志如下所示：
Aug 23 16:25:24 crick kernel: [617040.416257] ata4.00: exception Emask 0x0 SAct 0x1f SErr 0x0 action 0x0
Aug 23 16:25:24 crick kernel: [617040.416260] ata4.00: irq_stat 0x40000008
Aug 23 16:25:24 crick kernel: [617040.416262] ata4.00: failed command: READ FPDMA QUEUED
Aug 23 16:25:24 crick kernel: [617040.416265] ata4.00: cmd 60/08:08:00:af:cc/00:00:9c:00:00/40 tag 1 ncq 4096 in
Aug 23 16:25:24 crick kernel: [617040.416265] res 41/40:08:00:af:cc/00:00:9c:00:00/00 Emask 0x409 (media error) <F>
Aug 23 16:25:24 crick kernel: [617040.416266] ata4.00: status: { DRDY ERR }
Aug 23 16:25:24 crick kernel: [617040.416267] ata4.00: error: { UNC }
Aug 23 16:25:24 crick kernel: [617040.417510] ata4.00: configured for UDMA/133
Aug 23 16:25:24 crick kernel: [617040.417527] sd 3:0:0:0: [sdd] Unhandled sense code
Aug 23 16:25:24 crick kernel: [617040.417528] sd 3:0:0:0: [sdd]
Aug 23 16:25:24 crick kernel: [617040.417529] Result: hostbyte=DID_OK driverbyte=DRIVER_SENSE
Aug 23 16:25:24 crick kernel: [617040.417530] sd 3:0:0:0: [sdd]
Aug 23 16:25:24 crick kernel: [617040.417531] Sense Key : Medium Error [current] [descriptor]
Aug 23 16:25:24 crick kernel: [617040.417533] Descriptor sense data with sense descriptors (in hex):
Aug 23 16:25:24 crick kernel: [617040.417534] 72 03 11 04 00 00 00 0c 00 0a 80 00 00 00 00 00
Aug 23 16:25:24 crick kernel: [617040.417538] 9c cc af 00
Aug 23 16:25:24 crick kernel: [617040.417540] sd 3:0:0:0: [sdd]
Aug 23 16:25:24 crick kernel: [617040.417541] Add. Sense: Unrecovered read error - auto reallocate failed
Aug 23 16:25:24 crick kernel: [617040.417542] sd 3:0:0:0: [sdd] CDB:
Aug 23 16:25:24 crick kernel: [617040.417543] Read(10): 28 00 9c cc af 00 00 00 08 00
Aug 23 16:25:24 crick kernel: [617040.417547] end_request: I/O error, dev sdd, sector 2630659840
Aug 23 16:25:24 crick kernel: [617040.417550] md/raid:md128: read error not correctable (sector 2630657792 on sdd1).
Aug 23 16:25:24 crick kernel: [617040.417552] md/raid:md128: Disk failure on sdd1, disabling device.
Aug 23 16:25:24 crick kernel: [617040.417552] md/raid:md128: Operation continuing on 2 devices.
Aug 23 16:25:24 crick kernel: [617040.417563] ata4: EH complete
Aug 23 16:25:25 crick kernel: [617040.455605] RAID conf printout:
Aug 23 16:25:25 crick kernel: [617040.455609] --- level:5 rd:4 wd:2
Aug 23 16:25:25 crick kernel: [617040.455610] disk 0, o:1, dev:sdc1
Aug 23 16:25:25 crick kernel: [617040.455611] disk 1, o:0, dev:sdd1
Aug 23 16:25:25 crick kernel: [617040.455612] disk 2, o:1, dev:sde1
Aug 23 16:25:25 crick kernel: [617040.489941] RAID conf printout:
Aug 23 16:25:25 crick kernel: [617040.489945] --- level:5 rd:4 wd:2
Aug 23 16:25:25 crick kernel: [617040.489947] disk 0, o:1, dev:sdc1
Aug 23 16:25:25 crick kernel: [617040.489948] disk 2, o:1, dev:sde1
A slew of:
Aug 23 16:25:25 crick kernel: [617040.539926] Buffer I/O error on device md128, logical block 986401023
with different block addresses then
Aug 23 16:25:25 crick kernel: [617040.539929] EXT4-fs warning (device md128): ext4_end_bio:248: I/O error writing to inode 42993727 (offset 11551637504 size 524288 starting block 986400896)
Aug 23 16:25:25 crick kernel: [617040.541690] JBD2: Detected IO errors while flushing file data on md128-8
Aug 23 16:25:25 crick kernel: [617040.541707] Aborting journal on device md128-8.
Aug 23 16:25:25 crick kernel: [617040.541720] EXT4-fs error (device md128) in ext4_free_blocks:4702: Journal has aborted
Aug 23 16:25:25 crick kernel: [617040.541727] Buffer I/O error on device md128, logical block 1098416128
Aug 23 16:25:25 crick kernel: [617040.541729] lost page write due to I/O error on md128
Aug 23 16:25:25 crick kernel: [617040.541734] Buffer I/O error on device md128, logical block 0
Aug 23 16:25:25 crick kernel: [617040.541736] lost page write due to I/O error on md128
Aug 23 16:25:25 crick kernel: [617040.541740] JBD2: Error -5 detected when updating journal superblock for md128-8.
Aug 23 16:25:25 crick kernel: [617040.541743] EXT4-fs (md128): delayed block allocation failed for inode 49152114 at logical offset 994912 with max blocks 2048 with error -30
Aug 23 16:25:25 crick kernel: [617040.541745] EXT4-fs (md128): This should not happen!! Data will be lost
Aug 23 16:25:25 crick kernel: [617040.541745]
Aug 23 16:25:25 crick kernel: [617040.541806] EXT4-fs (md128): previous I/O error to superblock detected
Aug 23 16:25:25 crick kernel: [617040.542518] EXT4-fs error (device md128) in ext4_da_writepages:2390: Journal has aborted
Aug 23 16:25:25 crick kernel: [617040.542526] JBD2: Detected IO errors while flushing file data on md128-8
Aug 23 16:25:25 crick kernel: [617040.542529] Buffer I/O error on device md128, logical block 0
Aug 23 16:25:25 crick kernel: [617040.542531] lost page write due to I/O error on md128
Aug 23 16:25:25 crick kernel: [617040.542798] EXT4-fs error (device md128): ext4_journal_start_sb:371: Detected aborted journal
Aug 23 16:25:25 crick kernel: [617040.542810] EXT4-fs (md128): Remounting filesystem read-only
Aug 23 16:25:25 crick kernel: [617040.542815] EXT4-fs (md128): previous I/O error to superblock detected
Aug 23 16:25:25 crick kernel: [617040.542835] EXT4-fs (md128): I/O error while writing superblock
Aug 23 16:25:25 crick kernel: [617040.542838] EXT4-fs (md128): ext4_da_writepages: jbd2_start: 15984 pages, ino 49152114; err -30
Aug 23 16:25:25 crick kernel: [617040.544887] EXT4-fs error (device md128): ext4_journal_start_sb:371: Detected aborted journal
当我使用磁盘实用程序或 gparted sdf 消失时，它甚至不再被检测到。因此 md128 被降级。我做的第一件事是通过磁盘实用程序卸载它，然后我尝试看看是否有其他东西可以找到它：
fdisk -l
Disk /dev/sda: 256.1 GB, 256060514304 bytes
255 heads, 63 sectors/track, 31130 cylinders, total 500118192 sectors
Units = sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes
Disk identifier: 0x000a88fe
Device Boot Start End Blocks Id System
/dev/sda1 406368256 500117503 46874624 fd Linux raid autodetect
/dev/sda2 2046 406368255 203183105 5 Extended
/dev/sda5 2048 406368255 203183104 fd Linux raid autodetect
Partition table entries are not in disk order
Disk /dev/md0: 96.0 GB, 95998181376 bytes
2 heads, 4 sectors/track, 23437056 cylinders, total 187496448 sectors
Units = sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 524288 bytes / 1048576 bytes
Disk identifier: 0x00000000
Disk /dev/md0 doesn't contain a valid partition table
Disk /dev/md1: 207.9 GB, 207925149696 bytes
2 heads, 4 sectors/track, 50762976 cylinders, total 406103808 sectors
Units = sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes
Disk identifier: 0x00000000
Disk /dev/md1 doesn't contain a valid partition table
Disk /dev/sdb: 256.1 GB, 256060514304 bytes
255 heads, 63 sectors/track, 31130 cylinders, total 500118192 sectors
Units = sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes
Disk identifier: 0x000b0740
Device Boot Start End Blocks Id System
/dev/sdb1 * 406368256 500117503 46874624 fd Linux raid autodetect
/dev/sdb2 2046 406368255 203183105 5 Extended
/dev/sdb5 2048 406368255 203183104 fd Linux raid autodetect
Partition table entries are not in disk order
WARNING: GPT (GUID Partition Table) detected on '/dev/sdc'! The util fdisk doesn't support GPT. Use GNU Parted.
Disk /dev/sdc: 3000.6 GB, 3000592982016 bytes
255 heads, 63 sectors/track, 364801 cylinders, total 5860533168 sectors
Units = sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 4096 bytes
I/O size (minimum/optimal): 4096 bytes / 4096 bytes
Disk identifier: 0x00000000
Device Boot Start End Blocks Id System
/dev/sdc1 1 4294967295 2147483647+ ee GPT
Partition 1 does not start on physical sector boundary.
Disk /dev/md128: 9001.4 GB, 9001370124288 bytes
2 heads, 4 sectors/track, -2097367168 cylinders, total 17580801024 sectors
Units = sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 4096 bytes
I/O size (minimum/optimal): 524288 bytes / 1572864 bytes
Disk identifier: 0x00000000
Disk /dev/md128 doesn't contain a valid partition table
WARNING: GPT (GUID Partition Table) detected on '/dev/sde'! The util fdisk doesn't support GPT. Use GNU Parted.
Disk /dev/sde: 3000.6 GB, 3000592982016 bytes
255 heads, 63 sectors/track, 364801 cylinders, total 5860533168 sectors
Units = sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 4096 bytes
I/O size (minimum/optimal): 4096 bytes / 4096 bytes
Disk identifier: 0x00000000
Device Boot Start End Blocks Id System
/dev/sde1 1 4294967295 2147483647+ ee GPT
Partition 1 does not start on physical sector boundary.
WARNING: GPT (GUID Partition Table) detected on '/dev/sdd'! The util fdisk doesn't support GPT. Use GNU Parted.
Disk /dev/sdd: 3000.6 GB, 3000592982016 bytes
255 heads, 63 sectors/track, 364801 cylinders, total 5860533168 sectors
Units = sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 4096 bytes
I/O size (minimum/optimal): 4096 bytes / 4096 bytes
Disk identifier: 0x00000000
Device Boot Start End Blocks Id System
/dev/sdd1 1 4294967295 2147483647+ ee GPT
Partition 1 does not start on physical sector boundary.
猫/etc/fstab
# <file system> <mount point> <type> <options> <dump> <pass>
proc /proc proc nodev,noexec,nosuid 0 0
# / was on /dev/md127 during installation
UUID=b63b7341-0b85-40ed-a67b-acfe4f65f563 / ext4 errors=remount-ro 0 1
# /genome was on /dev/md128 during installation
UUID=bd6b54be-12ca-479d-879e-8d788fa9d039 /genome ext4 defaults 0 2
# swap was on /dev/md126 during installation
UUID=0470550a-6e92-485d-ad41-665c3f313287 none swap sw 0 0
~# cat /proc/mdstat
Personalities : [linear] [multipath] [raid0] [raid1] [raid6] [raid5] [raid4] [raid10]
md128 : active raid5 sde1[2] sdc1[0]
8790400512 blocks super 1.2 level 5, 512k chunk, algorithm 2 [4/2] [U_U_]
md1 : active raid1 sdb5[1] sda5[0]
203051904 blocks super 1.2 [2/2] [UU]
md0 : active raid0 sdb1[1] sda1[0]
93748224 blocks super 1.2 512k chunks
unused devices: <none>
~# mdadm --detail --scan
ARRAY /dev/md/0 metadata=1.2 name=crick:0 UUID=d0d08eab:a7e54021:25973acb:10dd5fba
ARRAY /dev/md/1 metadata=1.2 name=crick:1 UUID=e2774945:5b3ee3eb:2ad9390f:35153b82
ARRAY /dev/md/128 metadata=1.2 name=crick:128 UUID=345cb755:0ae1c919:d98a45ca:1baf3364
不过，这就是我对 Linux 的理解不够深入的地方。据我所知，阵列超级块可以看到它所基于的 sdf 的 UUID，但（无论出于什么原因）sdf 未被检测到。这意味着 RAID 阵列已损坏，因此 ubuntu 将文件系统设置为 RO 作为安全预防措施。我仍然可以打开、编辑和创建 / 上的文件，没有任何问题，操作系统仍然运行正常，我只是无法访问 md128 上的数据。我还没有关闭机器来移除驱动器，看看它是否发生故障。
大约一周前，同样的问题（随机切换到 RO）也出现了，只不过那次问题影响到了 md1（SSD）。发生这种情况时，md128 上的超级块被删除了。我重新安装操作系统时很幸运，它使用现有分区重建了 RAID（即没有格式化分区），并成功恢复了 99.9% 的数据（发生这种情况时正在访问的文件已损坏）。
- md128 的当前问题是否是我迄今为止设法解决的问题，还是我应该担心一些更深层次问题的征兆？
- 从这里继续前进的最佳方法是什么：替换 sdf 并重新条带化为 md128，替换 sdf 并拆除/重建 md128，还是其他什么？
编辑：抱歉，原帖者，我是论坛新手，没有读过格式说明，现在我已经改正了
答案1
很可能 sdf 前段时间坏了，而你却没有注意到。现在 sdd 有一些坏扇区。读取失败导致文件系统切换为只读。你需要更换故障驱动器，然后对文件系统进行 fsck。