From: Peter Van Oene (pvo@xxxxxxxxxxxx)
Date: Tue May 29 2001 - 12:31:11 GMT-3
Quick note here. OSPF network statements simply identify interfaces to be adde
d to the OSPF routing process. Unlike distance vector algorithms, they do not
signify networks to be advertised by the router. This is by design as link sta
te protocols advertise the status and configuration of links, not networks with
in the scope of a single area. Many other protocol implementations do not even
use this indirect method of adding interfaces to the process and configuration
involves specifying the interface name, not an IP range.
I think the better question is why did Cisco chose a method that involves a com
paring an address/mask to the set of configured interfaces in order to glean a
list of interfaces to participate in OSPF?. Why not just specify the interface
s outright?
Pete
*********** REPLY SEPARATOR ***********
On 5/29/2001 at 10:53 AM Walter Chen wrote:
>Jim, I'm glad that you tried it out instead of just talking about it with
>"reasoning". IMHO, Cisco IOS is a man-made product and worse, it is a
>piece
>of software! So by default it's not perfect and in many cases cannot be
>reasoned with. This includes all the standards too. I don't know this is
>a
>bug, but my simple taste is that I never liked to use 0.0.0.0 mask. It
>beats me why OSPF or Cisco IOS ever allowed to use a host mask to announce
>a
>network for a good "reason". To create confusion and chaos? Anyone knows?
>People talk about this should not make any difference, I think it got to
>make some difference. Why not?! Is there anywhere else in IP you can use
>different masks that do not make a difference? If not, why OSPF is so
>special?
>
>Walter
>
>-----Original Message-----
>From: Jim Graves [mailto:jtg@lucent.com]
>Sent: Tuesday, May 29, 2001 9:23 AM
>To: Brian Hescock; Jeff K.
>Cc: Mike Hess; ccielab@groupstudy.com
>Subject: Re: 0.0.0.0 wildcard bits and redistribution
>
>
>I can't think of any reason a 0.0.0.0 wildcard mask should make a
>difference either -- but it does. Give it a try. Last night, I set up
>the
>situation Walter talks about, and to my shock and horror he was right. I
>used a 0.0.0.0 wildcard mask, and the network didn't show up. When I
>changed the network statement to use a 0.0.0.255 wildcard mask, the
>network
>showed up in EIGRP. Huh.
>
>The same thing happens in bootcamp lab 3 if you replace the EIGRP LAN
>network with some other network outside 137.20.0.0/16. I tried it with
>both 192.168.50.0/24 and 172.20.50.0/24 (to rule out any kind of classful
>silliness). Again, when I used 0.0.0.0, the connected network didn't show
>up. With 0.0.0.255, it did. The same thing happened when I tried RIP v2
>or a second OSPF process instead of EIGRP.
>
>Why? Beats the heck out of me. I tested this using IOS versions
>12.0(11),
>12.0(15), and 12.0(7)T. I don't know if it's a bug of a feature, but it
>sure is weird.
>
>Jim
>
>At 10:35 PM 5/28/2001 -0400, Brian Hescock wrote:
>>I can't think of any way using a 0.0.0.0 wildcard bits would affect
>>redistribution, unless it's a bug. All the ospf network command does is
>>turn on ospf on
>>the interface(s) covered by the network statement, that's all it does. It
>>has zero effect on the network mask being advertised and no effect on
>>redistribution. To everyone that said use a 0.0.0.0 mask (wildcard bits
>>actually), thumbs up, it's the preferred method unless you have a lot of
>>interfaces you need to turn
>>ospf on. Not just for lab use but for production networks as well. Less
>>chance for error and easier to troubleshoot.
>>
>>Brian
>>
>>On Mon, 28 May 2001, Jeff K. wrote:
>>
>> > You are exactly right with all the shortcuts. I've never considered
>using
>> > aliases because I never use them day to day and know that it will wind
>up
>> > slowing me down since I type pretty fast. I know that I can calculate
>a
>> > wildcard mask easily and don't have a problem with it. Not my point,
>> > though... I was merely asking for an explanation as to why using the
>> > 0.0.0.0 area mask would affect route redistribution, which is a
>question
>> > that hasn't been answered. In my opinion, it makes good common sense
>> as you
>> > put it to use the 0.0.0.0 mask except when using a wildcard mask that
>will
>> > allow you to group multiple interfaces into a single area (i.e., a
>single
>> > statement versus multiple statements). Since I always use logic when
>> > assigning my interface addressing schemes, the all 0 mask allows me to
>> > double check everything quickly and easily - I know what interface has
>what
>> > address and where it should be. Not that the wildcard mask makes that
>any
>> > more difficult - just my personal preference. If a bug or other
>> > 'undocumented feature' requires me to use the exact mask, I will be
>fine
>as
>> > well... Anyway, if you have an answer for my original question about
>why
>> > the area mask affects route redistribution (what I originally
>responded
>> to),
>> > please respond. I am more curious than anything.
>> >
>> > Thanks,
>> >
>> > -Jeff
>> > ----- Original Message -----
>> > From: "Mike Hess" <mahess@home.com>
>> > To: <ccielab@groupstudy.com>
>> > Sent: Monday, May 28, 2001 7:06 PM
>> > Subject: RE: how anal is the lab grading
>> >
>> >
>> > > Why not just be smart and use a network statement and the right
>wildcard
>> > > mask for every interface that you are putting into OSPF? It seems
>common
>> > > sense to me not to try and get too pretty or cut too many corners
>and
>> then
>> > > get end up with a problem with 15 minutes left just because you were
>too
>> > > lazy to be more specific in the first place.
>> > >
>> > > Some people advocate the use of too many shortcuts. I see a lot of
>people
>> > > advocating shortcuts such as this, using a whold slew of alias
>commands,
>> > > etc. Just learn the material and don't try to save seconds when it's
>the
>> > > minutes and hours that really matter.
>> > >
>> > > Back to my original point, if you cannot at this point calculate a
>simple
>> > > wildcard mask then you are going for the wrong certification. Perhaps
>it
>> > is
>> > > not the grading.... :-)
>> > >
>> > > -----Original Message-----
>> > > From: nobody@groupstudy.com [mailto:nobody@groupstudy.com]On Behalf
>Of
>> > > Jeff K.
>> > > Sent: Monday, May 28, 2001 5:27 PM
>> > > To: Walter Chen; Peter Van Oene; ccielab@groupstudy.com
>> > > Subject: Re: how anal is the lab grading
>> > >
>> > >
>> > > You lost me on the second paragraph. Why would using the 0.0.0.0
>> > wild-card
>> > > mask for interface area identification affect redistribution? This
>mask
>> > > just allows you to use the interface's address. Obviously the
>subnet
>> mask
>> > > of your interfaces will affect redistribution, but I don't see how
>the
>> > > 0.0.0.0 area mask will. You can definitely mess up your OSPF
>topology
>by
>> > > using the wrong mask (i.e., interfaces in the wrong area, interfaces
>> added
>> > > to OSPF that weren't supposed to be). Let me know what your thoughts
>are
>> > on
>> > > this. Maybe I am forgetting something or am just misunderstanding
>> > > something.
>> > >
>> > > Thanks,
>> > >
>> > > -Jeff
>> > > ----- Original Message -----
>> > > From: "Walter Chen" <wchen@iloka.com>
>> > > To: "Peter Van Oene" <pvo@usermail.com>; <ccielab@groupstudy.com>
>> > > Sent: Monday, May 28, 2001 3:04 PM
>> > > Subject: Re: how anal is the lab grading
>> > >
>> > >
>> > > > IMHO, you can use 0.0.0.255 mask if you have other 1.1.1.x/29
>networks
>> > > onthe
>> > > > same router and they also belong to the same OSPF area. However,
>if
>> > these
>> > > > networks should not be in OSPF or in the same area, you definitely
>> > should
>> > > > not use 0.0.0.255 mask. The absolutely sure and correct way is to
>use
>> > > > 0.0.0.7 mask for 1.1.1.0/29 network.
>> > > >
>> > > > You can use 0.0.0.0 mask in cases if you do not need to
>redistribute
>> > > > 1.1.1.0/29 into other routing protocols on this same router.
>However,
>> > if,
>> > > > say, your 192.168.1.0/24 belongs to EIGRP and you want to
>reistribute
>> > > > between OSPF and EIGRP, the 1.1.1.0/29 network will NOT be passed
>into
>> > > EIGRP
>> > > > if you have used 0.0.0.0 mask.
>> > > >
>> > > > Walter
>> > > >
>> > > > ----- Original Message -----
>> > > > From: Peter Van Oene <pvo@usermail.com>
>> > > > To: <ccielab@groupstudy.com>
>> > > > Sent: Monday, May 28, 2001 9:35 AM
>> > > > Subject: Re: how anal is the lab grading
>> > > >
>> > > >
>> > > > > Personally, I'd not use 0.0.0.255 in any case, lab or otherwise.
>The
>> > > full
>> > > > 0's mask is the safe and accurate way to add interfaces to the OSPF
>> > > process
>> > > > and unless you need to add 20 odd interfaces, I'd suggest you use
>it.
>> > As
>> > > > far as the lab goes, I can attest that lab grading is fair. You
>> > shouldn't
>> > > > worry about trivial semantics. If your prepared, you'll likely
>have
>a
>> > > good
>> > > > idea when you are using an illegal shortcut.
>> > > > >
>> > > > > Pete
>> > > > >
>> > > > >
>> > > > > *********** REPLY SEPARATOR ***********
>> > > > >
>> > > > > On 5/28/2001 at 7:48 AM Don Dettmore wrote:
>> > > > >
>> > > > > >Question - How nit-picky are the graders of the lab? F0r
>example,
>> > > > > >something
>> > > > > >occurred to me when I was working in the lab:
>> > > > > >
>> > > > > >192.168.1.0 /24 ------ R1 ----- 1.1.1.0 /29
>> > > > > >
>> > > > > >When configuring R1 for OSPF, would the following be acceptable:
>> > > > > >
>> > > > > >network 1.1.1.0 0.0.0.255 area 0.0.0.0
>> > > > > >
>> > > > > >Or would that be considered wrong because of the 'wrong' (or I
>> should
>> > > > say -
>> > > > > >not specific enough) wildcard mask.
>> > > > > >
>> > > > > >Just wondering how anal I must train myself to be.
>> > > > > >
>> > > > > >Don Dettmore
>> > > > > >**Please read:http://www.groupstudy.com/list/posting.html
>> > > > > **Please read:http://www.groupstudy.com/list/posting.html
>> > > > **Please read:http://www.groupstudy.com/list/posting.html
>> > > **Please read:http://www.groupstudy.com/list/posting.html
>> > > **Please read:http://www.groupstudy.com/list/posting.html
>> > **Please read:http://www.groupstudy.com/list/posting.html
>>**Please read:http://www.groupstudy.com/list/posting.html
>**Please read:http://www.groupstudy.com/list/posting.html
>**Please read:http://www.groupstudy.com/list/posting.html
h
**Please read:http://www.groupstudy.com/list/posting.html
This archive was generated by hypermail 2.1.4 : Thu Jun 13 2002 - 10:30:56 GMT-3