acroman 发表于 2007-06-21 19:12

Disk and File-system Management on Solaris


                                ** Install a new hard disk   --> perform a reconfiguration boot
    # touch /reconfiguration
   ( creates a file allowing the OS to check for the presence of new components during startup)
    # sync; init 0    # prtconf- to view the system's device configuration info
** Disks & Slices -- 8 slices/disk in total ( 0 ~ 7 ); defined by offset: begin cylinder and the size    slice 0 is always the root slice (boot sector and root file system)    slice 1 - swap, slice 2 - backup    no cylinder overlap allowed!    disk label - first sector of a disk, also called the volume table of contents(VTOC)      -> contains disk's controller and geometry info, and partition table    boot block - sectors 2 to 16 of a disk,bootstrap program (bootblk)      only the root file system has an active boot block    super block - sectors 17 to 33 of a disk, a table that describes the file system      num of data blocks and cylinder groups, block size, name of mount points, state flag    cylinder groups - from sector 34 to end, data      -> allow a hard disk head to read more info in each pass ( improve disk access)                the 1st component in each sylinder group is a backup of the superblock for a disk      cylinder group block - an info table: data blocks/directories/inodesin the group      data blocks - 8 KB by default (eight fragments of 1024 bytes)            UFS uses fragments to store small files    logical device name:** View partition table    prtvtoc /dev/rdsk/c3t0d1s3    -h, -f, -s, -t, -m** Format   - to create disk slices, display the partition table, test and repair disks, etc      disk partitioning - disk, type, partition, current      wu - readonly, wm -writable      label - write slice configuration to the disk's label    disk diagnosis and repair - format, repair(SCSI), analyze, defect(SCSI)    disk label management - label, backup, verify, save, volname    ! - to execute commands from the Format utility    example:      >format   >partition   >0               # for root slice   tag, wm, 0 (start cylinder), 1499e (end cylinder)   >label   >quit   >verify    ** UNIX file system(UFS)    - default Solaris file system    - state flags (clean, stable, active, unknown)      unknown state prompts Solaris to run a file system check    - extended fundamental types (EFT), to provide 32-bit ID numbers for users, groups, and devices    - support for large file systems, up to 1TB (but Solaris Volume Manager is required in practice)    - support for large files, larger than 2 GB    UFS logging -> ensures consistency and reduces the need to run file system checks    (UFS creates a log of the changes that a data transaction involves, and applies changes for complete transactions only at system reboot.)      mount with -o logging option** Create file system    considerations: I/O workloadanddata size    newfs [-Nv] device_name** Mount file system   mount [-o options]      -r   readonly    -m   without making an entry in the /etc/mnttab    -g   globally in a cluster    -O   over an existing file system on the same mount point    -a   a list of mount points to a series of file systems in parallel    logging/nologging, nolargefiles/largefiles, noatime, setuid/nosuid    mountall      - mount all file systems listed in the vfstab /etc/vfstab    -l   only local file systems    umount   -f   force to umount   umountall    fuser [-c|-d] [-k]   (to obtain a list of processes currently accessing a file system)    -k   kill any running processes** fsck    ~ can detect and repair    - lost files and directories --> lost+found    - damaged superblocks --> a backup superblock in any of the cylinder groups    - corrupted inodes --> /lost+found    - bad data blocks --> /lost+found    - inconsistent cylinder blocks   fsck [-m|V|y|Y|n|N] [-o file_system_options] file_system      -m   check a file system    -V   view all options selected by default for this command    -y/Y answer yes to every question    -n/N answer no to every question    b=nspecify a backup superblock for a file system to use    c    convert the format of a file system    f    force to check even with a bad superblock    p    "preen" mode checking, no user interavention allowed    w    check write-only file systems** du   ~ determine disk usage for files or directories      -a   display the num of data blocks    -s   the total num of data blocks    -k   display the output in KB    -d   cannot cross file system boundaries    -o   exclude subdir    -r   include hidden files** quot    - how much space different users' files consume    -a   for all mounted file systems    -c   three columns (size specific)    -f   (user specific)* ff    ~ view a list of file names and inode statistics for a file system    ff          -mn, -an, -cn

               
               
               
               

本文来自ChinaUnix博客,如果查看原文请点:http://blog.chinaunix.net/u/16112/showart_325979.html
页: [1]
查看完整版本: Disk and File-system Management on Solaris