How much do you need to know about disks to successfully manage a Linux system? What commands do what? How do you make good decisions about partitioning? What kind of troubleshooting tools are available? What kind of problems might you run into? This article covers a lot of territory – from looking into the basics of a Linux file systems to sampling some very useful commands.
Disk technology
In the beginning days of Unix and later Linux, disks were physically large, but very small in terms of storage capacity. A 300 megabyte disk in the mid-90’s was the size of a shoebox. Today, you can get multi-terrabyte disks that are the size of a slice of toast.
Traditionally, files resided within file systems that resided in disk partitions that were themselves simply slices of disks. This organization still dominates today, though servers in large data centers often take on an entirely different structure.
/\
/ \
/ \
/ file \
/ \
/==========\
/ \
/ file system \
/ \
/==================\
/ disk partition \
/======================\
/ disk \
/==========================\
This simplistic view still works for many systems, but these days there are lot of complexities that make disk management harder in some ways and easier in others. A file system might be virtual – no longer residing on a single disk and more complex to manage, but far easier to resize as needed. In fact, the entire system could be virtual. And what we might manage as if it were a single disk could actually be some portion of a very large disk array.
Disk management
Sysadmins generally have to deal with many issues when it comes to managing disks. These include:
- Partitioning disks
- Creating file systems
- Mounting file systems
- Sharing file systems
- Monitoring free space within file systems
- Backing up (and sometimes restoring) file systems
The reasons to partition a disk include:
- protecting some file systems from running out of space (e.g., you may want the OS partition to be separated from home directories or applications to keep it from being affected if users’ files begin to take up far an excessive amount of disk space)
- improving performance
- allocating swap space
- facilitating maintenance and backups (e.g., you might be able to unmount /apps if it’s not part of / and you might want to back up /home more frequently than /usr)
- more efficient (and targeted) fsck
- maintaining (particularly on test systems) multiple operating systems
- reserving enough disk space for file system expansion
- sharing select file systems with other systems
Partitioning commands
For most Linux servers, partitioning is done before the servers are deployed. On the other hand, you might add disks at some later time or hold back some significant amount of free disk space for future use.
To make changes or verify partitions, enter a command such as fdisk /dev/sda to start fdisk interactively and then type m to see a list of the things that you can do with the fdisk command.
$ sudo fdisk /dev/sda Command (m for help): m Command action a toggle a bootable flag b edit bsd disklabel c toggle the dos compatibility flag d delete a partition l list known partition types m print this menu n add a new partition o create a new empty DOS partition table p print the partition table q quit without saving changes s create a new empty Sun disklabel t change a partition's system id u change display/entry units v verify the partition table w write table to disk and exit x extra functionality (experts only)
As you can see, the fdisk command provides a lot of functionality. The partitions that you set up may look something like this configuration in which four partitions have been set up on a single disk – /dev/sda.
sda +------------+------------------------+--------------------+------+ | / 40G | /home 80G | /apps 70G | swap | +------------+------------------------+--------------------+------+ sda1 sda2 sda3 sda4
Examining disk space and disk partitions
There are a number of excellent commands for examining disk partitions. The df command is one of the most commonly used commands for reporting on disk space usage. With the -h option, the df command displays the measurements in the most "human-friendly" format and that is, in fact, what the “h” is meant to imply. As you can see in the example below, the measurements are displayed in kilobytes, megabytes or gigabytes depending on the sizes rather than all using the same scale.
$ df -h Filesystem Size Used Avail Use% Mounted on udev 969M 4.0K 969M 1% /dev tmpfs 196M 1.1M 195M 1% /run /dev/sda1 37G 4.5G 31G 13% / none 4.0K 0 4.0K 0% /sys/fs/cgroup none 5.0M 0 5.0M 0% /run/lock none 980M 152K 979M 1% /run/shm none 100M 36K 100M 1% /run/user /dev/sda3 28G 44M 26G 1% /apps
The pydf command (think "python df" as it's really a python script) also provides a very useful disk usage display showing mount points and cute little illustrations for how full each partition is.
$ pydf Filesystem Size Used Avail Use% Mounted on /dev/sda1 37G 4534M 30G 12.1 [##...........] / /dev/sda3 27G 44M 26G 0.2 [.............] /apps
The parted command displays partition information in a different format:
$ sudo parted -l Model: ATA WDC WD800AAJS-60 (scsi) Disk /dev/sda: 80.0GB Sector size (logical/physical): 512B/512B Partition Table: msdos Number Start End Size Type File system Flags 1 1049kB 40.0GB 40.0GB primary ext4 boot 2 40.0GB 50.0GB 10.0GB primary linux-swap(v1) 3 50.0GB 80.0GB 30.0GB primary ext4
The lsblk (list block devices) command illustrates the relationship between disks and their partitions graphically and also supplies the major and minor device numbers and mount points.
$ lsblk NAME MAJ:MIN RM SIZE RO TYPE MOUNTPOINT sda 8:0 0 74.5G 0 disk ├─sda1 8:1 0 37.3G 0 part / ├─sda2 8:2 0 9.3G 0 part [SWAP] └─sda3 8:3 0 28G 0 part /apps
The fdisk command reports more details on disk partitions and uses very different numbers. You can also use fdisk to create or delete partitions, list unpartitioned space, change a partition type, or verify the partition table.
$ sudo fdisk -l Disk /dev/sda: 80.0 GB, 80026361856 bytes 255 heads, 63 sectors/track, 9729 cylinders, total 156301488 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: 0x000f114b Device Boot Start End Blocks Id System /dev/sda1 * 2048 78125055 39061504 83 Linux /dev/sda2 78125056 97656831 9765888 82 Linux swap / Solaris /dev/sda3 97656832 156301311 29322240 83 Linux
The sfdisk command is similar to fdisk, but makes some partition manipulation activities easier to perform.
$ sudo sfdisk -l -uM Disk /dev/sda: 9729 cylinders, 255 heads, 63 sectors/track Units = mebibytes of 1048576 bytes, blocks of 1024 bytes, counting from 0 Device Boot Start End MiB #blocks Id System /dev/sda1 * 1 38146 38146 39061504 83 Linux /dev/sda2 38147 47683 9537 9765888 82 Linux swap / Solaris /dev/sda3 47684 76318 28635 29322240 83 Linux /dev/sda4 0 - 0 0 0 Empty
NOTE: A mebibyte (MiB) = 220 bytes or 1,048,576 bytes.
The cfdisk command can also be used to display or manipulate disk partitions.
$ sudo cfdisk
cfdisk (util-linux 2.20.1)
Disk Drive: /dev/sda
Size: 80026361856 bytes, 80.0 GB
Heads: 255 Sectors per Track: 63 Cylinders: 9729
Name Flags Part Type FS Type [Label] Size (MB)
--------------------------------------------------------------------------
Pri/Log Free Space 1.05*
sda1 Boot Primary ext4 39998.99*
sda2 Primary swap 10000.27*
sda3 Primary ext4 30025.98*
Pri/Log Free Space 0.10*
[ Help ] [ New ] [ Print ] [ Quit ] [ Units ]
[ Write ]
Create new partition from free space
Monitoring disk performance
The iostat command can display statistics that illustrate how disks are performing, including how heavily they are being used. It also displays important measurements that show how busy the CPU is and how much of its resources are used for types of work. The system described below is idle more then 95% of the time. More importantly for our focus on disks, the %iowait (CPU waiting on disk IO) is very low. This would not be true if the disk were unusually busy and disk IO were a bottleneck.
$ iostat -x 60
Linux 3.13.0-129-generic (stinkbug) 08/31/2017 _x86_64_ (2 CPU)
avg-cpu: %user %nice %system %iowait %steal %idle
0.93 1.15 0.35 1.86 0.00 95.73
Device: rrqm/s wrqm/s r/s w/s rkB/s wkB/s avgrq-sz avgqu-sz await r_await w_await svctm %util
sda 8.37 3.26 13.41 2.79 341.14 191.82 65.79 0.61 37.60 30.40 72.14 2.52 4.08
Probably one of the most informative commands for looking at disk health is smartctl (part of smartmontools). While the command generates a lot of output, it provides valuable measurements that might help you pinpoint disk problems, particularly once you get used to working with its extensive output.