Linux Add a Swap File – HowTo

Posted on in Categories , , , , , last updated May 3, 2017

I need additional swap space to improve my Linux server/desktop system performance. How do I add a swap file to Linux system using command line options without creating a new partitions?

In Linux, as in most other Unix-like operating systems, it is common to use a whole partition of a hard disk for swapping. However, with the 2.6 Linux kernel, swap files are just as fast as swap partitions, although I recommend using a swap partition. The administrative flexibility of swap files outweighs that of partitions; since modern high capacity hard drives can remap physical sectors, no partition is guaranteed to be contiguous. You can add swap file as a dedicated partition or use following instructions to create a swap file.

Procedure To Add a Swap File Under Linux

You need to use the dd command to create swap file. The mkswap command is used to set up a Linux swap area on a device or in a file.

Step #1: Login as the Root User

Open a terminal window (select Applications > Accessories > Terminal) or login to remote server using the ssh client. Switch to the root user by typing su - (or sudo -s) and entering the root password, when prompted:
$ su -
OR
$ sudo -s

Step #2: Create Storage File

Type the following command to create 512MB swap file (1024 * 512MB = 524288 block size):
# dd if=/dev/zero of=/swapfile1 bs=1024 count=524288
Sample outputs:

524288+0 records in
524288+0 records out
536870912 bytes (537 MB) copied, 3.23347 s, 166 MB/s

Where,

  1. if=/dev/zero : Read from /dev/zero file. /dev/zero is a special file in that provides as many null characters to build storage file called /swapfile1.
  2. of=/swapfile1 : Read from /dev/zero write storage file to /swapfile1.
  3. bs=1024 : Read and write 1024 BYTES bytes at a time.
  4. count=524288 : Copy only 523288 BLOCKS input blocks.

Step #3: Secure swap file

Setup correct file permission for security reasons, enter:
# chown root:root /swapfile1
# chmod 0600 /swapfile1
A world-readable swap file is a huge local vulnerability. The above commands make sure only root user can read and write to the file.

Step #4: Set up a Linux swap area

Type the following command to set up a Linux swap area in a file:
# mkswap /swapfile1
Sample outputs:

Setting up swapspace version 1, size = 524284 KiB
no label, UUID=0e5e7c60-bbba-4089-a76c-2bb29c0f0839

Step #5: Enabling the swap file

Finally, activate /swapfile1 swap space immediately, enter:
# swapon /swapfile1

Step #6: Update /etc/fstab file

To activate /swapfile1 after Linux system reboot, add entry to /etc/fstab file. Open this file using a text editor such as vi:
# vi /etc/fstab
Append the following line:
/swapfile1 none swap sw 0 0
Save and close the file. Next time Linux comes up after reboot, it enables the new swap file for you automatically.

How do I verify swap is activated or not?

Simply use the free command:
$ free -m

             total       used       free     shared    buffers     cached
Mem:          1876       1798         77          0        119       1440
-/+ buffers/cache:        237       1638
Swap:         4607          0       4607

How can I display swap usage summary on Linux?

Type the following swapon command:
# swapon -s
Sample outputs:

Filename				Type		Size	Used	Priority
/dev/sda6                               partition	4194296	0	0
/swapfile1                              file		524280	0	-1

Another option is to view /proc/meminfo file:
$ less /proc/meminfo
$ grep -i --color swap /proc/meminfo
Sample outputs:

SwapCached:        30748 kB
SwapTotal:       6291448 kB
SwapFree:        6154008 kB

You can also use top command, atop command, and/or htop command to display information about swap usage:
# top
# atop
# htop
Sample outputs from a database server running on a CentOS Linux server:

  1  [||                        3.9%]     Tasks: 171, 106 thr; 1 running
  2  [                          0.0%]     Load average: 0.06 0.12 0.09 
  3  [||                        2.0%]     Uptime: 22 days, 07:07:28
  4  [                          0.0%]
  5  [                          0.0%]
  6  [                          0.0%]
  7  [                          0.0%]
  8  [                          0.0%]
  Mem[||||||||||||||||||2112/11909MB]
  Swp[|                   134/6143MB]

  PID USER      PRI  NI  VIRT   RES   SHR S CPU% MEM%   TIME+  Command
 8523 mysql      20   0 4545M  414M  4816 S  0.0  3.5  0:00.00 /usr/libexec/mysq
 8524 mysql      20   0 4545M  414M  4816 S  0.0  3.5  0:02.74 /usr/libexec/mysq
 8525 mysql      20   0 4545M  414M  4816 S  0.0  3.5  0:16.17 /usr/libexec/mysq
 8526 mysql      20   0 4545M  414M  4816 S  0.0  3.5  0:06.33 /usr/libexec/mysq
 8528 mysql      20   0 4545M  414M  4816 S  0.0  3.5  4:02.14 /usr/libexec/mysq
 8529 mysql      20   0 4545M  414M  4816 S  0.0  3.5  5:22.00 /usr/libexec/mysq
 8530 mysql      20   0 4545M  414M  4816 S  0.0  3.5  0:04.63 /usr/libexec/mysq
 8531 mysql      20   0 4545M  414M  4816 S  0.0  3.5  0:50.95 /usr/libexec/mysq
 8532 mysql      20   0 4545M  414M  4816 S  0.0  3.5  0:08.10 /usr/libexec/mysq
 9359 mysql      20   0 4545M  414M  4816 S  0.0  3.5 22:53.28 /usr/libexec/mysq

How can I disable devices and files for paging and swapping on Linux?

You need to use the swapoff command:
# swapoff /swapfile1
# swapon -s

How do I set swappiness on a Linux server?

The syntax is:
# sysctl vm.swappiness=VALUE
# sysctl vm.swappiness=20
OR
# echo VALUE > /proc/sys/vm/swappiness
# echo 30 > /proc/sys/vm/swappiness

The value in /proc/sys/vm/swappiness file controls how aggressively the kernel will swap memory pages. Higher values increase agressiveness, lower values descrease aggressiveness. The default value is 60. To make changes permanent add the following line to /etc/sysctl.conf:

echo 'vm.swappiness=30' >> /etc/sysctl.conf
See also:
This entry is 1 of 2 in the Linux and UNIX Swap File Management Tutorial series. Keep reading the rest of the series:
  1. Linux Add a Swap File
  2. FreeBSD Add a Swap File

Share this on:

51 comment

  3. If you don’t want to reboot the machine to enable the new swapfile, after step f) you can issue these commands:

    swapoff -a
    swapon -a

    to first stop and then start all swaps in /etc/fstab

    Reply

    1. this was really helpful for debugging my automounting issue. Turns out I had a typo on the swapfile name.
      Thnx some_guy

      Reply

  4. Thanks for the help. But how would I go about deleting this swapfile? Also, how does creating a swapfile this way differ from creating a separate partition for one? I only ask because I would like to have multiple installs of linux on my system, and I would like them to share the same swapfile.

    Recently, I’ve been trying to create a swapfile partition and it’s not being recognized, so I’ve been having trouble.

    Reply

  6. Thank you for the guide; I’d like to add another detail. In this line one may choose to use other units to make things more simple:

    # dd if=/dev/zero of=/swapfile1 bs=1024 count=524288

    You can turn it into:

    # dd if=/dev/zero of=/swapfile1 bs=1M count=512

    This means that the block size is 1 MB, so count=512 means “I need 512 megs”, there is no need to do any other calculations.

    Reply

  7. hi,
    i’m about to install ubuntu karmic on a new computer with thre sata hard drives.

    I usually allocate a little more of current ram in swap partition (5 gigas) so I can safely hibernate, but on a post I read in FreeBSD forums there’s a link to FreeBSD manual where states there should be a swap partition for every disk, not just one swap for the whole system.

    So, I know Linux is not *BSD but I wonder if the same is applicable here because installing a swap of 5 gigs in all three sata seems a waste of space to me!

    cheers!

    Reply

  8. Set swap priority:

    • Swapiness is the priority of input/output for swap. To look the current value:
    cat /proc/sys/vm/swappiness

    To change the swap priority (lower value means less swapping):
    sysctl vm.swappiness=10

    To have this value set at boot add it to /etc/sysctl.conf
    vm.swappiness=0

    Reply

  11. Great howto!

    After creating the swap file its permissions should be set so that only root can access the file:

    chmod 600 /swapfile1

    Reply

  14. So, after following these steps I received the notification that my root partition is full. And then, after rebooting, I cannot log into gnome. It will start gdm but will not go into gnome from there, it will only bring me back to gdm. startx does not work as well. So how do I do the oposite of this?

    Reply

  15. You just have to:
    -make sure the swap file is not active
    swapon -s
    -if it is active
    swapoff /swapfile (or whatever you called it)
    -comment out the line in your /etc/fstab file (Place a hashmark # in front of it or just delete the line)
    -remove the swapfile:
    rm /swapfile (or whatever you called it)

    Reboot and you should be away unless you want to create a smaller swap file in which case you should just make sure it isn’t in use, delete the file and recreate it using a smaller size.

    Reply

  16. Since kernel 2.6.31, you can use the util fallocate instead of dd on btrfs, ext4, ocfs2, and xfs filesystems. It’s *much* faster than dd on really big swap files.

    Reply

  18. dd if=/dev/zero of=/swapfile1 bs=1024 count=524288
    pls tell me the meaning and function of this command..why we are using if=/dev/zero and of=/swapfile..why this command started with dd..

    Reply

    1. dd is a command that does a low level copy of data from a to b. ‘if’ is the input file for the copy. /dev/zero is a device that generates as many null characters as it is asked to create and passes them on. ‘of’ is your output file or what will become your new (or additional) swapfile. the ‘/swapfile’ could be any name that you want it to be as long as you don’t forget what it is for the further manipulations. ‘bs’ is the block size of the file and ‘count’ is the number of bytes in the file. So, what the command does is takes 524288 null bytes from /dev/zero and puts them into /swapfile thus initially populating the file.

      Reply

    1. either create a new pagefile that’s larger mount it as swap, then umount the old swap and delete. then modify fstab as appropriate.

      or if it’s a lightly loaded machine you can unmount the swap create a larger swap and remount. This has the benefit that you can reuse the name ;-)

      Reply

    2. or…Of course you can create a 1gig file and just add that as swap as well. The only problem here is ensuring that the file is contiguous,

      i.e. 1x 3gb swap or 3x 1gb swap (swapfile1+swapfile2+swapfile3) or indeed 1x 2gb swap + 1x1gb swap.

      Reply

    1. You did not really say where you confusion lies. You have one of two options. You can either add a second swap file with whatever size will bring the total up to 2 GB or you can delete the current swapfile and establish a new one that is 2 GB in size. I believe both scenarios are covered in the comments above as well as the initial article so you will have to be a little more explicit if you want help …

      Reply

  23. We tried to add a swap file on one server and now using above steps but now it is growing big.

    [email protected] [/]# df -h
    Filesystem Size Used Avail Use% Mounted on
    /dev/sda2 4.9G 3.9G 710M 85% /
    tmpfs 1.9G 0 1.9G 0% /dev/shm
    /dev/sda1 485M 86M 374M 19% /boot
    /dev/sda6 176G 31G 136G 19% /home
    /dev/sda5 20G 6.6G 12G 36% /usr
    /dev/sda3 29G 6.3G 22G 23% /var
    /usr/tmpDSK 485M 16M 444M 4% /tmp
    [email protected] [/]# free -m
    total used free shared buffers cached
    Mem: 3810 3545 264 0 425 2288
    -/+ buffers/cache: 831 2979
    Swap: 2999 325 2674
    [email protected] [/]# grep swap /etc/fstab
    /swapfile swap swap defaults 0 0
    [email protected] [/]# du -sch /swapfile
    3.0G /swapfile
    3.0G total
    [email protected] [/]#

    Any solution ?

    Reply

  26. Thanks alot
    I was wondering to create my swap space due to out of space for oracle installation
    so the precheck was failed
    this gave me the correct result to over come my problem
    no need to restart the OS, it will apply after creating the swap file
    again very big thank for the support list

    Reply

  29. Unless something else has an express need to access /swapfile1 by a block device mount point “swap” this presents MORE of a security risk NOT LESS. The second ‘/etc/fstab’ field should be “none”. Consider ‘/media/sda1/pagefile.sys none swap defaults 0 0’.
    You have used the example ‘/swapfile1 swap swap defaults 0 0’. During init, otherwise unmounted TRUE block devices REQUIRE mounting “swap”–designated in the second field ‘/dev/sda3 swap auto defaults 0 0’. A generic swap-file (not on the root device) requires patching: that it’s device has been previously mounted in some rc script ‘mount /dev/sda1’ (assuming valid entry in /etc/fstab) along with an existing contiguous /swapfile1 (else create–use dd), ‘delay 1’ (must wait for completion), ‘mkswap /mnt/sda1/swapfile1’ must be rw (defaults) and compatible permissions for the device file system, and of course ‘swapon /mnt/sda1/swapfile1’.

    Reply

  30. Thank you very much. I moved wordpress to AWS ec2 and started getting Database establishing error. I found that out of 1.7GB only 30-40 mb was left. Now with these steps I created additional 512 mb. Hope now I do not get any memory issue.

    Reply

  31. A much faster way to create a 1GB swapfile is…

    dd if=/dev/zero of=/swapfile bs=1024 count=1 seek=1M

    This skips to the end of the swapfile and writes just one block leaving a giant “hole” that will be read as zeros. Disk space is allocated as the file is written to. Beware, you are allowed to create a swap file larger than you have free disk space.

    Reply

  32. Nuts, ignore my last message. If you try to turn on swap using the above you will get…

    # swapon /swapfile
    swapon: /swapfile: skipping – it appears to have holes.

    Reply

  33. You can use simple way

    # dd if=/dev/zero of=/opt/swapfile bs=1M count=1024

    1. block size is 1 MB
    2. count=1024 means 1GB (swap file Size)
    3. /opt/swapfile (location as you want swap file)
    then
    # mkswap /opt/swapfile
    Do entry in Fstab
    # swanon -a

    Reply

    1. #Like?#
      swapon /swapfile
      swapoff -a
      dd if=/dev/zero of=/opt/swapfile bs=1M count=5700
      mkswap /opt/swapfile
      swanon -a
      sysctl vm.swappiness=100
      swapon /swapfile
      #

      Reply

  34. swapon -s is useful in showing how much swap space and on what drives.

    Normally I would prefer to make a swap partition rather then making a swap file. That way your swap space does not crash if your filesystem does. (ie if / has an issue, at least your swap memory is safer…. )

    Very helpful article.

    Reply

    1. Huh? Why do you think that you have to remove it? If it is a swap file, you really likely don’t want to remove it unless you want to constantly run out of memory …

      If you really do want to remove it, read the posts, everything is covered that I can think of including how to remove swap files.

      Reply

  39. swapon /swapfile1 should NOT be
    one should edit /etc/fstab first then issue:
    swapon -a
    swap already up won’t be affected and this will also test if /etc/fstab is properly configured for swap.

    small thing I know, but is a nice optimization :P

    Reply

  42. Very nice post…

    I have one issue regarding the swapfile in Linux

    I am able to create the swapfile successfully and i ran the commands

    swapon -a

    swapfile is on, but the swap file is unused everytime i check

    when i check through the cmd “swapon -s” everytime i see that system is unable to use the swapfile.

    ************:/ # swapon -s
    Filename Type Size Used Priority
    /swapfile file 100663292 0 -1

    Any idea that why it is unable to use the swapfile.

    Any help is highly appreciated.
    Thanks in advance…

    Regards,
    Akash

    Reply

Leave a Comment

Tagged as: Tags , , , , , , ,