Ytdyklly

I have a bash script which checks if a swap exists or not and if it doesn't exists, then it creates one;

if free | awk '/^Swap:/ {exit !$2}'; then

    echo "Have swap, skipping"

    sudo swapoff -a

    sudo dd if=/dev/zero of=/swapfile bs=1M count=2048

    sudo mkswap /swapfile

    sudo swapon /swapfile

else

    fallocate -l 2G /swapfile

    chmod 600 /swapfile

    mkswap /swapfile

    swapon /swapfile

    sudo echo -e "/swapfile none swap sw 0 0 n" >> /etc/fstab

fi

Now I also want to add this a check to find out the swap size - if it already exists - because if the swap already exists and it is 4GB then I will downgrade it to 2GB instead of skipping.

How can I do this?

Edit: The script currently creates a 2GB of swap file whether it exists or not so if I have 4GB swap file, it will change it to 2GB but if I also have 2GB swap file it will still re-do the swap with 2GB. I thought this is not a good option to go with so that's why I wonder if I should add a swap size check or not?

Assuming your kernel is not really old and supports the fallocate system call (available since version 2.6.23, see man fallocate(1) and man fallocate(2)), fallocate will likely be fast, since it does not write data blocks. Hence, there is no big issue with always creating a new swap file. The only step you will likely want to perform conditionally is editing your fstab.

Assuming you either have no swap at all or have exactly one swap file whose path is /swapfile:

swapfile="/swapfile"



# Make sure swap is on

swapon --all



# Check if our assumptions hold

if    [ "$(swapon --show --noheadings | wc -l)" -gt 1 ]         ||

  (   [ "$(swapon --show --noheadings | wc -l)" -eq 1 ]         &&

    ( [ "$(swapon --show=TYPE --noheadings)" != 'file' ]        ||

      [ "$(swapon --show=NAME --noheadings)" != "$swapfile" ]

    )

  );  then

    echo "Unsafe to proceed, exiting."

    exit

fi



# Edit /etc/fstab if our file is not already there

if ! grep -q '^[[:blank:]]*'"$swapfile"'[[:blank:]]{1,}none[[:blank:]]{1,}swap[[:blank:]]{1,}' /etc/fstab;

then

    printf '%sn' "$swapfile none swap sw 0 0" >> /etc/fstab

fi



# Create/replace the swap file

swapoff --all

[ -f "$swapfile" ] && rm -- "$swapfile"

fallocate -l 2GiB -- "$swapfile"

chmod 600 "$swapfile"

mkswap -- "$swapfile"

swapon --all

You may still want to avoid unnecessarily turning off and on your swap: it may be a slow operation if a substantial fraction of it is used and it can have unwanted consequences if not enough memory is available.

To partially address these concerns (and also answer your original question), the last part of the above code can be enclosed in a conditional block:

# Check if we want to shrink the swap file i.e. it is bigger than 2 GiB

# (or if we have no swap file)

if  [ "$(free | awk '/^Swap:/ { print $2 }')" = "0" ] ||

    [ "$(free --bytes | awk '/^Swap:/ { print $2 }')" -gt 2147483648 ]; then



    # Create/replace the swap file

    # Same as above...



fi

Finally, assuming your kernel version is at least 3.14 and provides MemAvailable in /proc/meminfo (whose value is reported as the available column by free, see man free(1)), you can also check if you have enough available memory before trying to turn off swap.

The last code snippet becomes:

# Do we have no swap or more swap than 2GB?

# If yes, do we have more available memory than used swap?

if  ( [ "$(free | awk '/^Swap:/ { print $2 }')" = "0" ] ||

      [ "$(free --bytes | awk '/^Swap:/ { print $2 }')" -gt 2147483648 ]

    ) &&

    [ "$(awk '/MemAvailable:/ { print $2 }' /proc/meminfo)" -gt "$(free | awk '/Swap:/ { print $3 }')" ];

then

    # Create/replace the swap file

    # Same as above...

fi