Dealing with a network knowledge troll!

Discussion in 'Linux Networking' started by Mike Lovell, Aug 24, 2012.

  1. Mike Lovell

    Mike Lovell Guest

    -----BEGIN PGP SIGNED MESSAGE-----
    Hash: SHA1

    Hello,

    I've got a guy who claims these:

    10.255.255.81/24
    192.168.255.81/24

    Are invalid unicast IP's, because any IP with a "255" in it *anywhere*
    is reserved for being a broadcast IP.

    To which he cites:
    http://www.comptechdoc.org/independent/networking/guide/netaddressing.html


    Now I immediately know this is wrong because you can use a "0" or "255"
    on the end of a host IP:
    http://en.wikipedia.org/wiki/IPv4#Addresses_ending_in_0_or_255


    And of course the RFC says it's fine to use a "255" in the 1st to 3rd
    octets. Nothing about this is reserved or makes the address
    specifically a broadcast one.


    And you can even find reachable hosts out there with "255" in them:


    host ns2.televork.ee
    ns2.televork.ee has address 81.21.255.1

    host www.lepton-line.jp
    www.lepton-line.jp has address 121.83.255.1


    So every router between me and them (for me to reach them) would have to
    be broken it allow this, if the other guy is right about them being
    invalid!



    Question:

    Am I correct (the IP's are valid) or is he (the IP's are invalid)? And why?

    -----BEGIN PGP SIGNATURE-----

    iJwEAQECAAYFAlA27/YACgkQRuP0ePfiZW5UFgP9GigWPzFK8IUUumH5+UI7vLPl
    jyboGFAj2a30te5PAh8FuUbvfNWNz1PTa1xS3JITgTk76Jc+qfShblqOBtZuVSgu
    LbXKoSOk9RZPC+98AvIN9EEMaATUIegUpALZrtElCKGr/woPG+EimYnlnmiE0oH7
    q2b4K0w2BjtZKbH/Bis=
    =9qi/
    -----END PGP SIGNATURE-----
     
    Mike Lovell, Aug 24, 2012
    #1
    1. Advertisements

  2. Mike Lovell

    David Brown Guest

    You are correct.

    The wikipedia article is as good an explanation as any.
     
    David Brown, Aug 24, 2012
    #2
    1. Advertisements

  3. Mike Lovell

    Jorgen Grahn Guest

    If you really have found it stated in the RFCs and the RFCs are not
    obsoleted by later ones, of course you are right and don't have to ask
    around.

    (Why stop at octets 1--3, by the way? Surely a.b.c.255 can be a valid
    unicast address too?)

    /Jorgen
     
    Jorgen Grahn, Aug 24, 2012
    #3
  4. You are right, he is wrong.
    In short, because nothing forbids them; it would be bizarre and
    arbitrary to do so.

    The focus on the dotted-quad notation is an enormous red flag that
    someone doesn't know what they're talking about. It's a notational
    convenience for humans, not a technical property. Interpret IPv4
    addresses as binary numbers and it should become clearer just how weird
    banning eight-ones strings at particular positions in the middle of them
    would be.
     
    Richard Kettlewell, Aug 24, 2012
    #4
  5. Mike Lovell

    Moe Trin Guest

    On 24 Aug 2012, in the Usenet newsgroup comp.os.linux.networking, in article
    Those are RFC1918 (private) addresses. Tell the idiot to set up a
    couple of boxes in those ranges and try it himself.

    1122 Requirements for Internet Hosts - Communication Layers. R.
    Braden, Ed.. October 1989. (Format: TXT=295992 bytes) (Updates
    RFC0793) (Updated by RFC1349, RFC4379, RFC5884, RFC6093,
    RFC6298, RFC6633) (Also STD0003) (Status: STANDARD)

    RFC1122 section 3.2.1.3 is poorly written, but sub section '(d)' and
    '(e)' limit it to the last address in a network/subnet (the -1) only.
    RFC1878

    1878 Variable Length Subnet Table For IPv4. T. Pummill, B. Manning.
    December 1995. (Format: TXT=19414 bytes) (Obsoletes RFC1860)
    (Status: HISTORIC)
    Depends on the network mask, but sure. A facility I worked at for five
    years used a 255.255.252.0 (0xff ff fc 00) mask, so three out of four
    x.y.z.255 addresses were "normal" host addresses (only a broadcast
    where 'z' divided evenly by 4). Thus,

    198.18.4.0 network address
    198.18.4.1 thru
    198.18.7.254 are host addresses
    198.18.7.255 the ONLY broadcast address of the lot

    That meant 198.18.4.255, 198.18.5.255 and 198.18.6.255 were hosts.

    Old guy
     
    Moe Trin, Aug 24, 2012
    #5
  6. Mike Lovell

    Mike Lovell Guest

    -----BEGIN PGP SIGNED MESSAGE-----
    Hash: SHA1

    The guy rejects the RFC!

    "I also find it amusing that the stupid **** is arguing with a
    J. Postel RFC that is 21 years old. :)"

    Then goes on to say he's worked on IT for 40 years, yata-yata-yata


    I thought I would take it to a networking group (not sure which is the
    most active network group) and check.
    No reason, just I mentioned octet 4 earlier up the post (because of the
    WIKI cite). So that's a given.

    -----BEGIN PGP SIGNATURE-----

    iJwEAQECAAYFAlA3o2UACgkQRuP0ePfiZW60ewQAzkG04X/n322MUyvss9qmGsjE
    qRoThAhiCCfDvn7/+S7whyxv0N+G25CdubyEhSS8/hWde4GGO8LQX7rTMfUSzGR3
    jn+rWNAqUHFFtauo6fGkyzIEJYtYaEQrU9LkXPXwuIhPQxV7gvVfhx5N+zo89o3N
    Hb4EOaW/bCVxHiwqNhw=
    =XFwG
    -----END PGP SIGNATURE-----
     
    Mike Lovell, Aug 24, 2012
    #6
  7. Mike Lovell

    telsar Guest

    -----BEGIN PGP SIGNED MESSAGE-----
    Hash: SHA1

    Well, it may say its valid for use and all, and a lot depends on the
    mask used. But, it would be an odd thing to do as most avoid this
    value. Using it may find some bugs scattered around in software.
    Arguing about specs like this is a bit silly, anyway. Its IPv6 time,
    lots of room for adventure especially concerning the conversion from
    IPv4. Should have some dazzling times ahead.

    PS: Anyone seeing a "255" would think its some kind of broadcast
    thingy, except us.
    - --
    Steal a little and go to jail, steal a lot and become King.
    -----BEGIN PGP SIGNATURE-----
    Version: GnuPG v2.0.17 (MingW32)
    Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

    iQEcBAEBAgAGBQJQN7VnAAoJEGKUFsau5FJ0bwAH/0ypLlx3NYWXoMeAkDG6vpOR
    NClxtY5FgLWrPADJaPg7zpQ0M//APImBmJozua5Y93R0m56FkpUS+k9ASs7avNoQ
    TrCowRbv6TfNwR8Nm4Fq2I7T1dnS3Y6JUM+NjFuyH/LhoJHYL2yNjLpORPIxP1//
    g87qC6lHbj2tslZDl+9qIA3B9IoETm8bCF+fbU1j0QqMQiv6HgAT3WaAM19BNke4
    kRkgN3r3hJG9A78Zh6xck56/IXfFj553wSFgueL5Da09zZWn/DafPL5W8O5Qs005
    BnXkpHdBreGgQvWfZ17LUw1lixB6Jr/b+Amw0gxlZQKzNm/8ybJ1ZXWXUmb8GO0=
    =hUNN
    -----END PGP SIGNATURE-----
     
    telsar, Aug 24, 2012
    #7
  8. David W. Hodgins, Aug 24, 2012
    #8
  9. Mike Lovell

    Moe Trin Guest

    On Fri, 24 Aug 2012, in the Usenet newsgroup comp.os.linux.networking,
    About a week ago, of the 118784 networks allocated or assigned by the
    five Regional Internet Registries, 600 had a '255' in the second or
    third octet. 5 of those 600 (128.255.0.0 being UofIowa, 170.255.0.0.
    192.12.255.0, 192.51.255.0 and 192.66.255.0) were assigned in the late
    1980s. I think someone might have noticed by now.
    About a week ago, there were 11619 IPv6 networks - and for perspective
    there were 11675 IPv4 networks at the end of 1991.
    Nearly all computer users wouldn't even recognize an IP address of
    either version 4 _or_ 6.

    Old guy
     
    Moe Trin, Aug 24, 2012
    #9
  10. Mike Lovell

    Jorgen Grahn Guest

    Ok, you don't want to convince yourself he's wrong, but see /him/
    backpedal on his claim. That can be difficult with some people.
    Ah, yes you did. Sorry.

    /Jorgen
     
    Jorgen Grahn, Aug 24, 2012
    #10
  11. Mike Lovell

    telsar Guest

    I stand by this. The world is full of half baked network software
    especially in older gear and code.
    I did not say it wasn't true. IPv4 addresses were always scarce.
    I can hardly wait until IPv4 is eradicated and the world can be
    comprised of static addresses as "the goddess" had intended.
    That is for sure. Most network folks in industry deal in the reserved
    ones anyway. 10.0.0.0 and 192.168.0.0

    The real addresses are in the world of the ISPs and folks running real
    stuff on the Internet. Far fewer of these folks.
     
    telsar, Aug 24, 2012
    #11
  12. Mike Lovell

    Jorgen Grahn Guest

    You snipped the part by Moe Trin you responded to:

    About a week ago, of the 118784 networks allocated or assigned by
    the five Regional Internet Registries, 600 had a '255' in the
    second or third octet. 5 of those 600 (128.255.0.0 being UofIowa,
    170.255.0.0. 192.12.255.0, 192.51.255.0 and 192.66.255.0) were
    assigned in the late 1980s. I think someone might have noticed by
    now.
    I assume you what you stand by is your statement "most avoid this
    value". Citation needed, please. As he writes, 255 in addresses is
    legal, /and/ widely used.
    /That/ I can agree with.

    /Jorgen
     
    Jorgen Grahn, Aug 24, 2012
    #12
  13. Did you have any specific software in mind, or are you just making this
    up?
     
    Richard Kettlewell, Aug 25, 2012
    #13
  14. Mike Lovell

    telsar Guest

    I am making it up. Its been like 15 or more years since I had a piece
    of network gear wig out over this. Seems like it was a Livingston
    portmaster. And another was a pre-release of the Total Control Hub from
    3com they were hoping we would use. I found that little jewel, and
    also, an issue with their frame relay implementation.

    Doesn't really matter anyway, if you have no respect for others opinions.
     
    telsar, Aug 25, 2012
    #14
  15. Mike Lovell

    telsar Guest

    On 8/24/2012 5:23 PM, Jorgen Grahn wrote:

    I give up. Its not even possible to agree with you and point at why
    someone would be mistaken due to common practices.

    It doesn't matter anyway.
     
    telsar, Aug 25, 2012
    #15
  16. Mike Lovell

    khaliken

    Joined:
    Oct 4, 2012
    Messages:
    3
    Likes Received:
    0
    Its wonderful..really your thread is really very good and I appreciate it. You gave so nice info which is very well and amazing. I really impressed by your thread.
     
    khaliken, Oct 4, 2012
    #16
    1. Advertisements

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.