Shulou(Shulou.com)06/03 Report--

Linux disk and file system management

CPU,Memory (RAM), iCompo

I/O:Disks,Ethtercard

Disk: persisting data

Interface type:

IDE (ata): parallel port, 133MB/s

SCSI: parallel port, Ultrascsl320320MB/s

SATA: serial port, 6gbps

SAS: serial port, 6gbps

USB: serial port, 480MB/s

Parallel port: multiple devices can be connected to the same cable

IDE: two, master, slave

SCSI:

Broadband: 16-1

Narrowband: 8-1

Iops:io per second

Serial port: one

Hard disk: mechanical hard disk, solid state hard disk:

Mechanical hard disk:

Track: track

Sector: sectors, 512bytes

Cylinder: cylinder: the same track of different disks

Partition based on cylinder

Average seek time:

5400rpm,7200rpm,10000rpm,15000rpm

Linux's philosophy: everything is a document:

Device type:

Block (block): random access, data exchange unit is "block"

Character (character): linear access, data exchange unit is "character"

Device files: FHS

/ dev

Device file: the driver associated with the device; the access type of the device

Device number:

Major: the main device number, which distinguishes the type of device; used to indicate the driver required by the device

Minor: secondary device number, which distinguishes different devices under the same type; it is the access entry for specific devices.

Mknod command:

Make block or character special files

Mknod [OPTION]... NAME TYPE [MAJOR MINOR]

-m MODE: access to the created device file

Device file name: ICANN

Disk:

IDE: / dev/hd [a Meiz]

For example: / dev/hda / dev/hdb

SCSI,SATA,USB,SAS:/dev/sa [a Murz]

Zoning:

/ dev/hda#

/ dev/hda1,...

Note: CentOS 6 and 7 systems identify the hard drive device as / dev/sd [a Murz] #

How the device is referenced:

Device file name

Volume label

UUID

Disk partitions: MBR,GPT

MBR:0 sector

Master Boot Record

It is divided into three parts:

446bytes:bootloader, program, boot program that starts the operating system

64bytes: partition table. Each 16bytes identifies one partition. There can only be 4 partitions in total.

Validity identification of 2bytes:MBR area; 55AA bit is valid; otherwise, it is invalid

Identification of primary and extended partitions: 1-4

Logical Partition: 5 +

When we want to add a hard drive to the system:

1: partition the disk to create a new available partition

2: format the partition to create a file system available to the system

3: verify the file system

4: on the Linux system, you need to create a mount point (directory) and mount it to

Logical partitioning: fdisk

1: view the partition information of the disk

Fdisk-l [- u] [device]; lists partition information on the specified disk device

2: manage Partition

Fdisk device

Fdisk provides an interactive interface to manage partitions with many subcommands for different management functions

All operations are done in memory and are not synchronized directly to disk until the w command is saved to common commands on disk:

N: create a new partition

D: delete existing partitions

T: modify the partition type

L: view all existing ID

W: save and exit

Q: do not save exit

M: view help information

P: show existing partition information

Note: for new partitions created on disk devices that are already partitioned and one of them has been mounted, the kernel may be created after the creation is complete

Cannot be directly identified.

View disk partitions: cat / proc/partitions

Notify the kernel to force rereading the disk partition table:

CentOS 5:partprobe [device]

CentOS 6,7:partx,kpartx

Increase partition synchronization

Partx-a [device]

Kpartx-af [device]

Delete partition synchronization

Partx-d-nr 7-8 [device]

Zone creation tools:

Parted,sfdisk

File system management tools:

Tools for creating file systems:

Mkfs

Mkfs.ext2,mkfs.ext3 mkfs.ext4 mkfs.ext4 mkfs.xfs mkfs.vfat

Tools for detecting and repairing file systems:

Fsck

Fsck.ext2 et al.

Tools for viewing its properties:

Dumpe2fs,tune2fs

Adjust file system properties:

Tune2fs

Components of the kernel level file system:

File system driver: provided by the kernel

File system management tools: provided by applications of user controls

Create a file system:

Formatting: low-level formatting (before partitioning, dividing tracks)

Advanced formatting (partition after partition, create file system)

Create a file system:

Linux file system type: ext2 ext3 ext4 xfs relserfs iso9660 swap

Metadata area

Metadata area:

Does not include file name

File metadata: inode (index node)

Size, permissions, belonging to master group, timestamp, block pointer, etc.

Symbolic link files: storing data

Management tools for ext series file systems:

Disk formatting:

Mkfs-t ext3=mkfs.ext3

In formatting to ext3, since we do not specify the specific options for the file system, the system uses default values for formatting.

Specify ext2/ext3 specific options (label,block size or number of inode)

Mke2fs [OPTIONS] device

-t: indicates the type of file system to be created

Mkfs-t ext3=mkfs.ext3

-b: {1024 | 2048 | 4096}: indicates the block size of the file system

-j: create a journaling file system ext3 (default is ext2)

Make2fs-j=make2fs-t ext3=mkfs-t ext3=mkfs.ext3

-I #: indicates the ratio of inode to bytes; that is, an inode is created for every number of bytes

-N #: directly indicates the number of inode to be created for this file system

-m #: specify reserved space, percentage

The volume is labeled: hello kitty

Block:2048

8192bytes assigns an inode

Build the ext3 file system of journal

E2label command: viewing and setting of volume label

Viewing: e2label device

Settings: e2label device LABEL

Tune2fs command: view or modify some properties of the ext series file system

Note: the block size cannot be modified after it is created

Tune2fs [OPTIONS] device

-l: view the contents of the super block

Modify the properties of the specified file system:

-j:ext2-> ext3

-L LABEL: modify the volume label

-m #: adjust the percentage of reserved space

-O [^] FEATHER: turn a feature on or off

-o [^] mount_options: turns some default mount option on or off

Acl

^ acl

Dumpe2fs command: displays attribute information for ext series file system

Dumpe2fs [- h] device

Fsck command: a tool for file system detection

File corruption may occur when the finalization operation is abnormally terminated due to unexpected termination of the process or system crash; at this time, the file system should be detected and repaired; it is recommended to proceed offline.

E2fsck [options] device

-y: automatically answer yes to all questions

-f: force detection even if the file system is in the clean state

Step 1: check inode, block, size

Step 2: check the directory structure

Step 3: check directory connectivity

Step 4: check the reference information

Step 5: check the cluster profile information

Fsck command:

-t fstype: indicates the file system type

Fsck-t ext4=fsck.ext4

-a: automatically fix all errors without interaction

-r: interactive repair

Blkid command: query the system type used on the device, UUID

Blkid device

Blkid-L LABEL: locate the device according to LABEL

Blkid-U UUID: locate the device according to UUID

Swap file system:

Swap partitions on Linux must use separate file systems

And the System ID of the file system must be 82

Create swap device: mkswap command

Mkswap [options] device

-L LABEL: indicates the volume label

-f: mandatory

Windows does not recognize Linux's file system; therefore, when a storage device needs to be cross-used between two systems, it should use a file system supported by both windows and Linux: fat32 (vfat)

# mkfs.vfat device

Use of the file system:

First mount: the mount command and the umount command

For a file system other than the root file system to be accessible, it must be "associated" to a directory on the file system for internship. This association operation is called "mount"; this directory is the "mount point".

Mount point: mount_point, which is used as an access entry to another file system

(1) exist beforehand

(2) directories that are not or will not be used by other processes should be used.

(3) the original files under the mount point will be hidden.

Mount command:

Mount [- nrw] [- t file system] [- L LABEL] [- o options] device dir

Command options:

-r:readonly, read-only mount

-w:read and write, read write mount

-n: by default, device mount or unmount operations are synchronously updated in / etc/mtab; used to disable this operation

-t vfstype: indicates the type of file system on the device to be mounted; in most cases it can be omitted

At this point, mount will determine the file type of the device to be mounted according to blkid.

-L LABEL: indicates the device as a volume label when mounting

-U UUID: indicates the device as UUID when mounting

-o options: mounting option

Sync/async: synchronous / asynchronous operations; (synchronous: write to memory and disk at the same time)

Atime/noatiome: whether to update the access timestamp when a file or directory is accessed

Diratime/nodiratime: whether to update the access timestamp when the directory is accessed

Remount: remount

Acl: supports the use of facl features

Example: no acl is added, and operation is not supported (setting access control function is not supported)

Only read and write permissions are found under mount

We use mount to remount and add the acl option

Enter / mnt again to see if it has access control function.

Acl acquiesce that we don't support it, it's too dangerous, you know, there's no need to dig a hole for yourself.

The second method:

Tune2fs-o acl / dev/sda6 enable acl

Tune2fs-l / dev/sda6 to see if acl is enabled

Ro: read-only

Rw: read and write

Dev/nodev: whether device files are allowed on this device

Exec/noexec: whether to allow running program files on this device

Auto/noauto: whether to allow ordinary users to mount this file system

Suid/nosuid: whether to allow suid and sgid special permissions on program files to take effect

Defaults: Use default options () default option):

Rw, suid, dev, exec, auto, nouser, async, and relatime

A technique to use:

You can bind a directory to another directory as a temporary access intersection

Mount-- bind source directory target directory

View all devices that have been attached to the current system

# mount

# cat / etc/mtab

# cat / proc/mounts

Mount the CD:

Mount-r / dev/cdrom mount_point (mount point)

Hang on the flash drive:

Identify the device file of the USB disk in advance

Mount the local loopback device:

Mount-o loop / PATH/TO/SOME_LOOP_FILE MOUNT_POINT

Umount command:

Umount device | dir

Note: the mount point being accessed by the process cannot be uninstalled

Check which process or processes are occupied by:

# lsof MOUNT_POINT or # fuser-v MOUNT_POINT

Terminate all processes that are accessing a mount point:

# fuser-km MOUNT_POINT forced T-drop

# unmount umount MOUNT_POINT again

Enable the swap partition:

Command to create a swap partition: mkswap

Enable: swapon

Swapon [OPTION] [DEVICE]

-a: all swap devices are defined in the / etc/fstab file

Disable: swapoff

Swapoff DEVICE

Set the file system other than the root file system to be automatically mounted when it is powered on: / etc/fstab file

Each line defines a file system to be mounted and related properties:

6 field

(1) the equipment to be hung:

Device file

LABEL

UUID

Pseudo file system: such as sysfs,proc,tmpfs, etc.

(2) Mount point

The mount point for devices of type swap is swap

(3) File system type

(4) Mount option

Defaults: use the default mount option

If you want to use multiple mount options at the same time, you can separate them by commas.

Example: defaults,acl,noatime,noexc

(5) dump frequency

0: never back up

1: backup every day

2: backup every other day

(6) self-check order

0: no self-test

1: self-test first, usually only the root file system is available 1

2: secondary self-test

...

Mount-a: all devices defined in this file that support auto-mount can be automatically mounted.

(1) automatically mount with LABEL

(2) automatically mount with device

Two commands: df and du

Df command:

Df [OPTION] [FILE]

-l: displays only relevant information about local files

-h:human-readable

-l: displays inode usage status instead of blocks

Du command:

Du [OPTION] [FILE]

-s:sumary

-h:human-readble

Lunix disk and file management system blogger summed up here, today is Sunday, bloggers began to sum up from the afternoon to writing a blog, intermittent in the middle, now there is a feeling of exhaustion! Then I'll see you in the next section. My wish is that there are no more witty bitches in the world.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.