Re: blocklistd.conf blocking an ipv6 /48 subnet

2025-12-30 Thread Brook Milligan


> On Dec 30, 2025, at 10:56, Greg Troxel  wrote:
> 
> I typed 'man blacklistd' and then 'man blacklistd.conf' on a NetBSD 9
> system.  On NetBSD 10 is blocklist and seems a little more detailed.
> Same on NetBSD 11.

OK, I had forgotten about the name change.  

> 
>>> [remote]
>>> 0.0.0.0/0   stream  tcp *   =/24=   =
>>> #[0::0]/0   stream  tcp *   =/64=   =
> 
> This is in /usr/share/examples/blacklistd/blacklistd.conf on 9 but the
> newer examples have dropped the v6 example and are generally more
> confusing.

I would have expected those files to come from 
https://cvsweb.netbsd.org/bsdweb.cgi/src/external/bsd/blacklist/ or more 
recently blocklist, but I still cannot find anything like that in either place. 
 I wonder where the files come from.
> 
>> This clears things up a lot.  Unless I missed something, perhaps this should 
>> be added to the cvs tree.
> 
> I think it's just a question of clarifying the examples and adding back
> a :: => 64 line.

Yes, that would be a big help.  (Once we know where the files are. :)

> I think it is prudent.  I just meant that after you experience it for a
> while, I am curious what you see.

Thanks for clarifying.  I’ll let it run and see what happens.

I mainly wanted to be sure that I understood how the ipv6 addresses where being 
handled / blocked.  Your notes help a lot.

Cheers,
Brook



Re: blocklistd.conf blocking an ipv6 /48 subnet

2025-12-30 Thread Greg Troxel
Brook Milligan  writes:

>> On Dec 30, 2025, at 09:45, Greg Troxel  wrote:
>> 
>> I am looking at the NetBSD 9 man pages and example, reproduced below
>
> I’m confused; where did you find those man pages?  I’m seeing no such 
> information in the cvs tree 
> (https://cvsweb.netbsd.org/bsdweb.cgi/src/external/bsd/blocklist/bin/), the 
> published man pages, or anywhere else.  Hence my question.

I typed 'man blacklistd' and then 'man blacklistd.conf' on a NetBSD 9
system.  On NetBSD 10 is blocklist and seems a little more detailed.
Same on NetBSD 11.

>> [remote]
>> 0.0.0.0/0   stream  tcp *   =/24=   =
>> #[0::0]/0   stream  tcp *   =/64=   =

This is in /usr/share/examples/blacklistd/blacklistd.conf on 9 but the
newer examples have dropped the v6 example and are generally more
confusing.

> This clears things up a lot.  Unless I missed something, perhaps this should 
> be added to the cvs tree.

I think it's just a question of clarifying the examples and adding back
a :: => 64 line.

>> With 9 (blacklistd), not having a remote entry for v6 leads to a /128
>> being blocked.  (In my experience this is super rare.)
>
> OK, so there may be no need for an extra ipv6 block, I suppose.

Depends on if your attackers are like my attackers.  Blocking a 64 on v6
bad behavior makes sense to me; it's like blocking a v4 /24, more or
less.

>> I can see why you want to block a /48, but would be interested if you
>> are willing to share the details of the kind of bad behavior you
>> experience, and if there is a pattern of blocking /64  and then later
>> having a failure form a later /64 within the same /48.
>
> I have no problems with ipv6 addresses, but wanted to block them as I
> do ipv4.  I figured the same approach (blocking subnets) would be
> prudent, but perhaps that is not necessary in practice.

I think it is prudent.  I just meant that after you experience it for a
while, I am curious what you see.

I see /48 as a prefix assigned to a site/etc. and /64 to an individual
"link" (e.g. ethernet).


Re: blocklistd.conf blocking an ipv6 /48 subnet

2025-12-30 Thread Brook Milligan


> On Dec 30, 2025, at 09:45, Greg Troxel  wrote:
> 
> I am looking at the NetBSD 9 man pages and example, reproduced below

I’m confused; where did you find those man pages?  I’m seeing no such 
information in the cvs tree 
(https://cvsweb.netbsd.org/bsdweb.cgi/src/external/bsd/blocklist/bin/), the 
published man pages, or anywhere else.  Hence my question.

> [remote]
> 0.0.0.0/0   stream  tcp *   =/24=   =
> #[0::0]/0   stream  tcp *   =/64=   =

This clears things up a lot.  Unless I missed something, perhaps this should be 
added to the cvs tree.

> With 9 (blacklistd), not having a remote entry for v6 leads to a /128
> being blocked.  (In my experience this is super rare.)

OK, so there may be no need for an extra ipv6 block, I suppose.

> I can see why you want to block a /48, but would be interested if you
> are willing to share the details of the kind of bad behavior you
> experience, and if there is a pattern of blocking /64  and then later
> having a failure form a later /64 within the same /48.

I have no problems with ipv6 addresses, but wanted to block them as I do ipv4.  
I figured the same approach (blocking subnets) would be prudent, but perhaps 
that is not necessary in practice.

Thanks for your help.

Cheers,
Brook



Re: blocklistd.conf blocking an ipv6 /48 subnet

2025-12-30 Thread Greg Troxel
Brook Milligan  writes:

> I understand that the blocklistd.conf syntax for blocking an ipv4 subnet is 
> the following:
>
>  [remote]
>  * stream tcp * =/24 = =
>
> Presumably, there is something similar for blocking an ipv6 subnet.  I
> have tried replacing the address (field 1) with [*] and the name
> (field 5) with =/48, but that seems to block ipv4 subnets with a mask
> of /48.
>
> What is the syntax for blocking a ipv6 /48 subnet?

I am looking at the NetBSD 9 man pages and example, reproduced below

[remote]
0.0.0.0/0   stream  tcp *   =/24=   =
#[0::0]/0   stream  tcp *   =/64=   =

I think you need to have a :: to cause the parser to treat the first
field as a v6 literal.

With 9 (blacklistd), not having a remote entry for v6 leads to a /128
being blocked.  (In my experience this is super rare.)

I can see why you want to block a /48, but would be interested if you
are willing to share the details of the kind of bad behavior you
experience, and if there is a pattern of blocking /64  and then later
having a failure form a later /64 within the same /48.