Looking at the XFS Filesystem On Linux
The XFS journaling filesystem is yet another filesystem originally created for a commercial Unix system that made its way into the Linux world. Silicon Graphics Incorporated (SGI) originally created XFS in 1994 for its commercial IRIX Unix system.
It was released to the Linux environment for common use in 2002. The XFS filesystem has recently become more popular and is used as the default filesystem in mainstream Linux distributions, such as RHEL.
The XFS filesystem uses the writeback mode of journaling, which provides high performance but does introduce an amount of risk because the actual data isn’t stored in the journal file. The XFS filesystem also allows online resizing of the filesystem, similar to the Reiser4 filesystem, except XFS filesystems can only be expanded and not shrunk.
For filesystems, an alternative to journaling is a technique called copy-on-write (COW).
COW offers both safety and performance via snapshots. For modifying data, a clone or writable-snapshot is used. Instead of writing modified data over current data, the modified data is put in a new filesystem location. Even when data modification is completed, the old data is never overwritten.
COW filesystems are gaining in popularity. Two of the most popular, Btrfs and ZFS, are briefly reviewed in the following sections.
ZFS is a stable filesystem and competes well against Resier4, Btrfs, and ext4. Its biggest detractor is that ZFS does not have a GPL license. The OpenZFS project was launched in 2013, which may help to change this situation. However, it’s possible that until a GPL license is obtained, ZFS will never be a default Linux filesystem.
The openSUSE Linux distribution recently established Btrfs as its default filesystem. It is also offered in other Linux distributions, such as RHEL, although not as the default filesystem.
The partition can be an entire disk or a subset of a disk that contains a portion of the virtual directory.
The fdisk utility is used to help you organize partitions on any storage device installed on the system. The fdisk command is an interactive program that allows you to enter commands to walk through the steps of partitioning a hard drive.
To start the fdisk command, you need to specify the device name of the storage device you want to partition and you need to have superuser privileges. When you don’t have superuser privileges and attempt to use fdisk, you’ll receive some sort of error message, like this one:
$ fdisk /dev/sdb
Unable to open /dev/sdb
$
If you do have superuser privileges and the correct device name, the fdisk command allows you entrance into the utility as demonstrated here on a CentOS distribution:
$ sudo fdisk /dev/sdb
[sudo] password for Christine:
Device contains neither a valid DOS partition table,
nor Sun, SGI or OSF disklabel
Building a new DOS disklabel with disk identifier 0xd3f759b5.
Changes will remain in memory only
until you decide to write them.
After that, of course, the previous content won’t be recoverable.
Warning: invalid flag 0x0000 of partition table 4 will
be corrected by w(rite)
[…]
Command (m for help):
The fdisk interactive command prompt uses single letter commands to instruct fdisk what to do. Table 8.2 shows the commands available at the fdisk command prompt.
Although this list may look intimidating, usually you need just a few basic commands in day-to-day work.
For starters, you can display the details of a storage device using the p command:
Command (m for help): p
Disk /dev/sdb: 5368 MB, 5368709120 bytes
255 heads, 63 sectors/track, 652 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes
Disk identifier: 0x11747e88
Device Boot Start End Blocks Id System
Command (m for help):
The output shows that the storage device has 5368MB of space on it (5GB). The listing under the storage device details shows whether there are any existing partitions on the device. The listing in this example doesn’t show any partitions, so the device is not partitioned yet.
Next, you’ll want to create a new partition on the storage device. Use the n command for that:
Command (m for help): n
Command action
e extended
p primary partition (1-4)
p Partition number (
1-4): 1
First cylinder (1-652, default 1): 1
Last cylinder, +cylinders or +size{K,M,G} (1-652, default 652): +2G
Command (m for help):
Partitions can be created as either a primary partition or an extended partition. Primary partitions can be formatted with a filesystem directly, whereas extended partitions can only contain other primary partitions. The reason for extended partitions is that there can only be four partitions on a single storage device. You can extend that by creating multiple extended partitions and then creating primary partitions inside the extended partitions.
This example creates a primary storage device, assigns it partition number 1, and then allocates 2GB of the storage device space to it. You can see the results using the p command again:
Command (m for help): p
Disk /dev/sdb: 5368 MB, 5368709120 bytes
255 heads, 63 sectors/track, 652 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes
Disk identifier: 0x029aa6af
Device Boot Start End Blocks Id System
/dev/sdb1 1 262 2104483+ 83 Linux
Command (m for help):
Now in the output there’s a partition shown on the storage device (called /dev/sdb1). The Id entry defines how Linux treats the partition. fdisk allows you to create lots of partition types. Using the l command lists the different types available. The default is type 83, which defines a Linux filesystem. If you want to create a partition for a different filesystem (such as a Windows NTFS partition), just select a different partition type.
You can repeat the process to allocate the remaining space on the storage device to another Linux partition. After you’ve created the partitions you want, use the w command to save the changes to the storage device:
Command (m for help): w
The partition table has been altered!
Calling ioctl() to re-read partition table.
Syncing disks.
$
The storage device partition information was written to the partition table, and Linux was informed of the new partition via the ioctl() call. Now that you have set up a partition on the storage device, you’re ready to format it with a Linux filesystem.
 |
| Looking at the XFS Filesystem On Linux |
It was released to the Linux environment for common use in 2002. The XFS filesystem has recently become more popular and is used as the default filesystem in mainstream Linux distributions, such as RHEL.
The XFS filesystem uses the writeback mode of journaling, which provides high performance but does introduce an amount of risk because the actual data isn’t stored in the journal file. The XFS filesystem also allows online resizing of the filesystem, similar to the Reiser4 filesystem, except XFS filesystems can only be expanded and not shrunk.
Understanding the copy-on-write filesystems
With journaling, you must choose between safety and performance. Although data mode journaling provides the highest safety, performance suffers because both inode and data is journaled. With writeback mode journaling, performance is acceptable, but safety is compromised.For filesystems, an alternative to journaling is a technique called copy-on-write (COW).
COW offers both safety and performance via snapshots. For modifying data, a clone or writable-snapshot is used. Instead of writing modified data over current data, the modified data is put in a new filesystem location. Even when data modification is completed, the old data is never overwritten.
COW filesystems are gaining in popularity. Two of the most popular, Btrfs and ZFS, are briefly reviewed in the following sections.
Looking at the ZFS Filesystem
The COW filesystem ZFS was developed in 2005 by Sun Microsystems for the OpenSolaris operating system. It began being ported to Linux in 2008 and was finally available for Linux production use in 2012.ZFS is a stable filesystem and competes well against Resier4, Btrfs, and ext4. Its biggest detractor is that ZFS does not have a GPL license. The OpenZFS project was launched in 2013, which may help to change this situation. However, it’s possible that until a GPL license is obtained, ZFS will never be a default Linux filesystem.
Looking at the Btrfs Filesystem
The COW newcomer is the Btrfs filesystem, also called the B-tree filesystem. Oracle started development on Btrfs in 2007. It was based on many of Reiser4’s features, but offered improvements in reliability. Additional developers eventually joined in and helped Btrfs quickly rise toward the top of the popular filesystems list. This popularity is due to stability, ease of use, as well as the ability to dynamically resize a mounted filesystem.The openSUSE Linux distribution recently established Btrfs as its default filesystem. It is also offered in other Linux distributions, such as RHEL, although not as the default filesystem.
Working with Filesystems
Linux provides a few different utilities that make it easier to work with filesystems from the command line. You can add new filesystems or change existing filesystems from the comfort of your own keyboard. This section walks you through the commands for interacting with filesystems from a command line environment.Creating partitions
To start out, you need to create a partition on the storage device to contain the filesystem.The partition can be an entire disk or a subset of a disk that contains a portion of the virtual directory.
The fdisk utility is used to help you organize partitions on any storage device installed on the system. The fdisk command is an interactive program that allows you to enter commands to walk through the steps of partitioning a hard drive.
To start the fdisk command, you need to specify the device name of the storage device you want to partition and you need to have superuser privileges. When you don’t have superuser privileges and attempt to use fdisk, you’ll receive some sort of error message, like this one:
$ fdisk /dev/sdb
Unable to open /dev/sdb
$
Note
Sometimes, the hardest part of creating a new disk partition is trying to find the physical disk on your Linux system. Linux uses a standard format for assigning device names to hard drives, but you need to be familiar with the format. For older IDE drives, Linux uses /dev/hdx, where x is a letter based on the order the drive is detected (a for the first drive, b for the second, and so on). For both the newer SATA drives and SCSI drives, Linux uses /dev/sdx, where x is a letter based on the order the drive is detected (again, a for the first drive, b for the second, and so on). It’s always a good idea to double-check to make sure you are referencing the correct drive before formatting the partition!
If you do have superuser privileges and the correct device name, the fdisk command allows you entrance into the utility as demonstrated here on a CentOS distribution:
$ sudo fdisk /dev/sdb
[sudo] password for Christine:
Device contains neither a valid DOS partition table,
nor Sun, SGI or OSF disklabel
Building a new DOS disklabel with disk identifier 0xd3f759b5.
Changes will remain in memory only
until you decide to write them.
After that, of course, the previous content won’t be recoverable.
Warning: invalid flag 0x0000 of partition table 4 will
be corrected by w(rite)
[…]
Command (m for help):
Tip
If this is the first time you’re partitioning the storage device, fdisk gives you a warning that a partition table is not on the device.
The fdisk interactive command prompt uses single letter commands to instruct fdisk what to do. Table 8.2 shows the commands available at the fdisk command prompt.
 |
| The fdisk Commands |
For starters, you can display the details of a storage device using the p command:
Command (m for help): p
Disk /dev/sdb: 5368 MB, 5368709120 bytes
255 heads, 63 sectors/track, 652 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes
Disk identifier: 0x11747e88
Device Boot Start End Blocks Id System
Command (m for help):
The output shows that the storage device has 5368MB of space on it (5GB). The listing under the storage device details shows whether there are any existing partitions on the device. The listing in this example doesn’t show any partitions, so the device is not partitioned yet.
Next, you’ll want to create a new partition on the storage device. Use the n command for that:
Command (m for help): n
Command action
e extended
p primary partition (1-4)
p Partition number (
1-4): 1
First cylinder (1-652, default 1): 1
Last cylinder, +cylinders or +size{K,M,G} (1-652, default 652): +2G
Command (m for help):
Partitions can be created as either a primary partition or an extended partition. Primary partitions can be formatted with a filesystem directly, whereas extended partitions can only contain other primary partitions. The reason for extended partitions is that there can only be four partitions on a single storage device. You can extend that by creating multiple extended partitions and then creating primary partitions inside the extended partitions.
This example creates a primary storage device, assigns it partition number 1, and then allocates 2GB of the storage device space to it. You can see the results using the p command again:
Command (m for help): p
Disk /dev/sdb: 5368 MB, 5368709120 bytes
255 heads, 63 sectors/track, 652 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes
Disk identifier: 0x029aa6af
Device Boot Start End Blocks Id System
/dev/sdb1 1 262 2104483+ 83 Linux
Command (m for help):
Now in the output there’s a partition shown on the storage device (called /dev/sdb1). The Id entry defines how Linux treats the partition. fdisk allows you to create lots of partition types. Using the l command lists the different types available. The default is type 83, which defines a Linux filesystem. If you want to create a partition for a different filesystem (such as a Windows NTFS partition), just select a different partition type.
You can repeat the process to allocate the remaining space on the storage device to another Linux partition. After you’ve created the partitions you want, use the w command to save the changes to the storage device:
Command (m for help): w
The partition table has been altered!
Calling ioctl() to re-read partition table.
Syncing disks.
$
The storage device partition information was written to the partition table, and Linux was informed of the new partition via the ioctl() call. Now that you have set up a partition on the storage device, you’re ready to format it with a Linux filesystem.
Tip
Some distributions and older distribution versions do not automatically inform your Linux system of a new partition after it is made. In this case, you need to use either the partprobe or hdparm command (see their man pages), or reboot your system so it reads the updated partition table
0 Response to "Looking at the XFS Filesystem On Linux"
Post a Comment