Report size of all Amazon S3 buckets with S3CMD

s3_bucket_size.pl

#!/usr/bin/perl
# I use updraft to backup WordPress sites, I need to monitor S3 bucket size because money
# This script will report the size used by all buckets, used to monitor the
# size of all buckets.
# It uses s3cmd https://s3tools.org/s3cmd
# Outputs and send via email the size of all buckets, sorted by size, with
# total size of all buckets
#
# Stephan Burlot [email protected] Apr 2024
#

use Data::Dumper;
use MIME::Lite;
use POSIX;

# change the config file here if needed
$s3cmd = "s3cmd --config=/home/stephan/.s3cfg ";
$email_address = '[email protected]';
# I use the same script to check backblaze/b2 buckets, so for now this is empty
$b2_bucket = "";

################################################################################################
# general routine to send results via email
sub send_email($) {

	my $message = shift @_;

    my $msg = MIME::Lite->new(
        From    => $email_address,
        To      => $email_address,
        Subject => 'Backups S3 Buckets Sizes',
        Data    => "Amazon S3 usage:\n\n$message\n\n"
    );
    $msg->send;
}

################################################################################################

## get list of buckets/folders
$cmd = $s3cmd . "ls " . $b2_bucket;
$response = qx/$cmd/;
my $status = "";

# extract name of buckets/folders from s3cmd response
# bucket/folder name is column 3
my @BUCKETS = ();
for $line (split("\n", $response)) {
	$bucket = (split(/\s+/, $line))[2];
	push @BUCKETS, $bucket;
}

# enumerate all buckets/folders and get the size
for $bucket_url (@BUCKETS) {
	$cmd = $s3cmd . " du $bucket_url";
	$response = qx/$cmd/;
	chomp($response);
	$response =~ s/^\s+//g;
	($bucket = $bucket_url) =~ s/$b2_bucket//;
	$bucket =~ s/\/$//;
	$size = (split(/\s+/, $response))[0];
	# we get the size in bytes and transform in M
	# there's no way to directly get the size in M
	$size = ceil($size/1048576) . "M";
	# that's not what I would call error checking
	if ($response eq "") {
		print( "Bucket: $bucket is empty\n" );
		next;
	}

	printf("Bucket: %-10s %s\n", $size, $bucket);
	$status .= sprintf("%s\t%s\n", $size, $bucket);
}

## get the total size of the bucket/folder. We could add all folder size, but I prefer to be sure
$total_cmd = $s3cmd . "du -H " . $b2_bucket;
$response = qx/$total_cmd/;
$response = (split("\n", $response))[-1];
$response =~ s/^\s+//g;
$size = (split(/\s+/, $response))[0];

# sort the bucket list by size
$cmd = "echo \"$status\" | sort -n";
$sorted = qx/$cmd/;

$sorted .= "--------------\n$size\tTotal\n";
print "\n\nSorted:\n$sorted\n";
# send the result via email
send_email($sorted);