Linux FSCK自动修复文件系统（按F键）

原创

wx63b7e0251fc35 2023-01-08 00:01:35 博主文章分类：Ros、Linux相关 ©著作权

©著作权归作者所有：来自51CTO博客作者wx63b7e0251fc35的原创作品，请联系作者获取转载授权，否则将追究法律责任

背景：

Linux系统（Ubuntu）在运行时，断电等非正常关机操作，会导致ext4文件系统数据损坏。严重时会导致系统崩溃。如下log就是系统数据损坏。

检查方法：

1. [    7.878756] EXT4-fs error (device mmcblk0p2): ext4_mb_generate_buddy:742: group 0, 14845 clusters in bitmap, 14822 in gd  
2. [    8.484660] init: samba-ad-dc main process (995) terminated with status 1  
3. [   14.248075] EXT4-fs error (device mmcblk0p2): ext4_mb_generate_buddy:742: group 1, 1 clusters in bitmap, 2 in gd

1、开机log，如上log就是开机时，kernel监测到文件系统错误；

2、比如要检查的分区是/dev/mmcblk0p2，如下红色字体部分就是系统错误的信息。

1. ~# tune2fs -l /dev/mmcblk0p2  
2. tune2fs 1.42.9 (4-Feb-2014)  
3. Filesystem volume name:   <none>  
4. Last mounted on:          /  
5. Filesystem UUID:          ab013911-6048-465f-8a1a-cf1420c7bb01  
6. Filesystem magic number:  0xEF53  
7. Filesystem revision #:    1 (dynamic)  
8. Filesystem features:      ext_attr resize_inode dir_index filetype extent flex_bg sparse_super large_file huge_file uninit_bg dir_nlink extra_isize  
9. Filesystem flags:         unsigned_directory_hash  
10. Default mount options:    user_xattr acl  
11. Filesystem state:         not clean with errors  
12. Errors behavior:          Continue  
13. Filesystem OS type:       Linux  
14. Inode count:              393216  
15. Block count:              1572864  
16. Reserved block count:     78643  
17. Free blocks:              289870  
18. Free inodes:              169990  
19. First block:              0  
20. Block size:               4096  
21. Fragment size:            4096  
22. Reserved GDT blocks:      383  
23. Blocks per group:         32768  
24. Fragments per group:      32768  
25. Inodes per group:         8192  
26. Inode blocks per group:   512  
27. Flex block group size:    16  
28. Filesystem created:       Sat Sep 12 11:55:02 2015  
29. Last mount time:          Mon Oct  5 00:56:20 2015  
30. Last write time:          Mon Oct  5 00:56:32 2015  
31. Mount count:              49  
32. Maximum mount count:      -1  
33. Last checked:             Sat Sep 12 11:55:02 2015  
34. Check interval:           0 (<none>)  
35. Lifetime writes:          5936 MB  
36. Reserved blocks uid:      0 (user root)  
37. Reserved blocks gid:      0 (group root)  
38. First inode:              11  
39. Inode size:               256  
40. Required extra isize:     28  
41. Desired extra isize:      28  
42. Default directory hash:   half_md4  
43. Directory Hash Seed:      37b6421d-4697-4ff5-a68e-4a1e1ea81c0e  
44. Journal backup:           inode blocks  
45. <span style="color:#ff0000;">FS Error count:           5  
46. First error time:         Mon Oct  5 00:52:47 2015  
47. First error function:     ext4_mb_generate_buddy  
48. First error line #:       742  
49. First error inode #:      0  
50. First error block #:      0  
51. Last error time:          Mon Oct  5 00:56:32 2015  
52. Last error function:      ext4_mb_generate_buddy  
53. Last error line #:        742  
54. Last error inode #:       0  
55. Last error block #:       0</span>

修复方法：

1、手动修复：借助其他完整系统启动，对所在磁盘分区卸载，比如要修复/dev/mmcblk0p2,

执行命令 fsck.ext4 /dev/mmcblk0p2 可检查修复系统；

2、自动修复：

条件：

（1）、

console=tty1 console=ttySAC2,115200n8 root=UUID=e139ce78-9841-40fe-8823-96a304a09859 rootwait ro

如果最后ro是rw，将不能完成自动修复。

（2）、

UUID=e139ce78-9841-40fe-8823-96a304a09859 / ext4 errors=remount-ro,noatime,nodiratime 01

（3）、

# automatically repair filesystems with inconsistencies during boot
FSCKFIX=yes

然后，可以参考/etc/init/mountall.conf

1. description     "Mount filesystems on boot"  
2.   
3. start on startup  
4. stop on starting rcS  
5.   
6. expect daemon  
7. task  
8.   
9. emits virtual-filesystems  
10. emits local-filesystems  
11. emits remote-filesystems  
12. emits all-swaps  
13. emits filesystem  
14. emits mounting  
15. emits mounted  
16.   
17. script  
18. default/rcS || true  
19. "--force-fsck"  
20. "$FSCKFIX" = "yes" ] && fsck_fix="--fsck-fix"  
21.   
22.     # Doesn't work so well if mountall is responsible for mounting /proc, heh.  
23. if [ -e /proc/cmdline ]; then  
24.         read line < /proc/cmdline  
25. for arg in $line; do  
26. case $arg in  
27.                 -q|--quiet|-v|--verbose|--debug)  
28.                     debug_arg=$arg  
29.                     ;;  
30.             esac  
31.         done < /proc/cmdline  
32.     fi  
33.     # set $LANG so that messages appearing in plymouth are translated  
34. if [ -r /etc/default/locale ]; then  
35. default/locale || true  
36.         export LANG LANGUAGE LC_MESSAGES LC_ALL  
37.     fi

1.     exec mountall --daemon $force_fsck $fsck_fix $debug_arg  
2. end script

1. post-stop script  
2. true  
3. end script

（4）、系统检测到分区有问题时，会再根目录下创建一个空文件/forcefsck，重启后，执行mountall，自动进行修复，然后删除forcefsck，也可以手动创建/forcefsck，系统同样会在下次启动时强制检查修复文件系统；

Log：

系统启动检查修复过程的log，不在/var/log/fsck/目录下，而是在/var/log/upstart/目录下，文件为 mountall.log，如下：

1. # cat mountall.log  
2. mount: mount point /media/boot does not exist  
3. mountall: mount /media/boot [382] terminated with status 32  
4. mountall: Filesystem could not be mounted: /media/boot  
5. Skipping /media/boot at user request  
6. Skipping /media/boot at user request  
7. Skipping /media/boot at user request  
8. fsck from util-linux 2.20.1  
9. e2fsck 1.42.9 (4-Feb-2014)  
10. /dev/mmcblk0p2: clean, 223220/393216 files, 1282976/1572864 blocks

其他：

也可以通过设置系统挂载的次数来自动检查修复文件系统

比如：

tune2fs -c 30 /dev/mmcblk0p2 系统每启动30次，就会检查修复一次。

修复完成后，通过 tune2fs -l /dev/mmcblk0p2看到没有错误信息，如下：

1. ~# tune2fs -l /dev/mmcblk0p2  
2. tune2fs 1.42.9 (4-Feb-2014)  
3. Filesystem volume name:   <none>  
4. Last mounted on:          /  
5. Filesystem UUID:          ab013911-6048-465f-8a1a-cf1420c7bb01  
6. Filesystem magic number:  0xEF53  
7. Filesystem revision #:    1 (dynamic)  
8. Filesystem features:      ext_attr resize_inode dir_index filetype extent flex_bg sparse_super large_file huge_file uninit_bg dir_nlink extra_isize  
9. Filesystem flags:         unsigned_directory_hash  
10. Default mount options:    user_xattr acl  
11. Filesystem state:         not clean  
12. Errors behavior:          Continue  
13. Filesystem OS type:       Linux  
14. Inode count:              393216  
15. Block count:              1572864  
16. Reserved block count:     78643  
17. Free blocks:              289888  
18. Free inodes:              169996  
19. First block:              0  
20. Block size:               4096  
21. Fragment size:            4096  
22. Reserved GDT blocks:      383  
23. Blocks per group:         32768  
24. Fragments per group:      32768  
25. Inodes per group:         8192  
26. Inode blocks per group:   512  
27. Flex block group size:    16  
28. Filesystem created:       Sat Sep 12 11:55:02 2015  
29. Last mount time:          Mon Oct  5 00:58:56 2015  
30. Last write time:          Mon Oct  5 00:58:41 2015  
31. Mount count:              1  
32. Maximum mount count:      -1  
33. Last checked:             Mon Oct  5 00:58:41 2015  
34. Check interval:           0 (<none>)  
35. Lifetime writes:          5938 MB  
36. Reserved blocks uid:      0 (user root)  
37. Reserved blocks gid:      0 (group root)  
38. First inode:              11  
39. Inode size:               256  
40. Required extra isize:     28  
41. Desired extra isize:      28  
42. Default directory hash:   half_md4  
43. Directory Hash Seed:      37b6421d-4697-4ff5-a68e-4a1e1ea81c0e  
44. Journal backup:           inode blocks