The way the node name encoding is implemented is problematic as we
operate outside of integer ranges:
[I] febner@dev8 ~/repos/pve/qemu-server (qemu-blockdev-options)> cat asdf.pm
use strict;
use warnings;
use Digest::SHA;
sub encode_base62 {
my ($input) = @_;
my @chars = ('0'..'9', 'A'..'Z', 'a'..'z');
my $base = 62;
my $value = 0;
my $result = '';
for my $byte (unpack('C*', $input)) {
$value = $value * 256 + $byte;
print "$value\n";
}
}
encode_base62(Digest::SHA::sha1("foo"));
[I] febner@dev8 ~/repos/pve/qemu-server (qemu-blockdev-options)> perl
asdf.pm
11
3054
782023
200198069
51250705898
13120180709951
3358766261747471
859844163007352795
2.20120105729882e+20
5.63507470668499e+22
1.44257912491136e+25
3.69300255977307e+27
9.45408655301907e+29
2.42024615757288e+32
6.19583016338658e+34
1.58613252182696e+37
4.06049925587703e+39
1.03948780950452e+42
2.66108879233157e+44
6.81238730836881e+46
Also, while the use case here shouldn't be cryptographically sensitive,
you never know, so I'll just use a different hash function than sha1.
I'll cut off the result from that hash to 30 hex digits. Then we still
have one letter for the prefix of the node name.
As for collision probability, that will be 120 bits and should be more
than enough, even 2^50 nodes have a very small probability to collide
with that:
>>> math.log2(16**30)
120.0
>>> d=2**120
>>> n=2**50
>>> 1 - math.exp(-(n*n)/(2*d))
4.768370445162873e-07
_______________________________________________
pve-devel mailing list
pve-devel@lists.proxmox.com
https://lists.proxmox.com/cgi-bin/mailman/listinfo/pve-devel
