Use another group to extract all the IP:port strings, and use split to extract them later.

qr/
        ^\s*
        (\S+)                                        ##  something
        \s+
        (\S+)                                        ##  apache24
        \s+
        (\S+)                                        ##  up
        \s+
        (\S+)                                            ##  11572
        \s+
        (.*?)                                            ##  date-time
        ((?:                                             ##  start non
+-capturing group for ports*
            \s+
            (?:                                          ##  PORTS
                (?:25[0-5]|2[0-4][0-9]|[0-1]?[0-9]{1,2})
                [.]
                (?:25[0-5]|2[0-4][0-9]|[0-1]?[0-9]{1,2})
                [.]
                (?:25[0-5]|2[0-4][0-9]|[0-1]?[0-9]{1,2})
                [.]
                (?:25[0-5]|2[0-4][0-9]|[0-1]?[0-9]{1,2})
                [:]
                \d+
            )
        )+)                                              ##  end non-c
+apturing group for ports*
        $
    /x;

my @ips = split ' ', $6;
[download]

Update: fixed the regex.

($q=q:Sq=~/;[c](.)(.)/;chr(-||-|5+lengthSq)`"S|oS2"`map{chr |+ord
}map{substrSq`S_+|`|}3E|-|`7**2-3:)=~y+S|`+$1,++print+eval$q,q,a,
[download]

I got a string that I want to analyse, preferably with a regexp.

This is just what split was made for. Start with the seventh element of the array returned by split and proceed until you reach the end of the array. No need to confuse the issue with a fancy and fragile Regex.....

my $line = 'somthing       apache24 up      11572    Mar 15 16:25 161.
+20.224.243:8808 161.20.224.243:8802 161.20.224.243:8809';


my ($col1,$col2,$col3,$size,$month,$date,$time,@ips) =
    split /\s+/, $line;
    
    
print "@ips";

__END__
161.20.224.243:8808 161.20.224.243:8802 161.20.224.243:8809
[download]

It is possible to write a regex that does exactly what the split would do. But since splitting on white space works so well in this example, I didn't bother to do any regex testing. The key regex feature that is applicable here is "match global". If you had trouble say getting everything to work as a single regex, there is nothing wrong with writing one to get the initial scalars, and then run a second regex to get all of the ip's.  @ips=/($ipregex)/g; That "g" at the end is what does the "match global" magic. In most cases, running a couple of regex's on the same line makes no performance problem.

Update: You didn't show your left hand assignment statement, but what I show above is what you want whether using split or a regex. Of course you should have better names than $col1, $col2, etc. In general I recommend not messing around with $1,$2 etc. Get these values into a readable name ASAP. I very seldom have any code like: $tokens[3]. In that case, you need a comment to describe what index 3 means... Far better is to have a name like $size which I demonstrate.