From: Bryan Bartik (bbartik@ipexpert.com)
Date: Sat Mar 14 2009 - 21:10:03 ARST
Yep, I see now that RFC 3031 says "An LSA with the P-bit set is preferred
over one with the P-bit" and if the P-bit settings are equal, the highest
router-id is preferred.
Here is the actual LSA itself, the NP is set to 1.
LS Type: NSSA AS-External-LSA
LS Age: 175 seconds
Do Not Age: False
Options: 0x28 (DC, NP)
0... .... = DN: DN-bit is NOT set
.0.. .... = O: O-bit is NOT set
..1. .... = DC: Demand Circuits are supported
...0 .... = L: The packet does NOT contain LLS data block
.... 1... = NP: NSSA is supported
.... .0.. = MC: NOT multicast capable
.... ..0. = E: NO ExternalRoutingCapability
Link-State Advertisement Type: NSSA AS-External-LSA (7)
Link State ID: 192.168.100.0
Advertising Router: 1.1.1.1 (1.1.1.1)
LS Sequence Number: 0x80000002
LS Checksum: 0x8845
Length: 36
Netmask: 255.255.255.0
External Type: Type 1 (metric is specified in the same units as
interface cost)
Metric: 20
Forwarding Address: 192.168.12.1
External Route Tag: 0
The p-bit from R4's LSA is of course clear. I thought maybe Cisco was
skipping the first step so I changed R1's router-id to something higher than
R4, but E1 was still preferred over N1. It wouldn't be the first time Cisco
decided to do things their own way :)
On Sat, Mar 14, 2009 at 4:41 PM, Scott Morris <swm@emanon.com> wrote:
> When you do a "show ip ospf database nssa (ip-route#)" do you see anything
> about the P-bit or not?
>
> If the p-bit is present I would expect the N1 to be chosen. If the p-bit
> is
> not present, I would expect the E1 to be chosen.
>
> RFC 3101 details the process that is supposed to happen.
>
> I'll try to get some time to lab that up though as it would be intriguing
> if
> not following expected behavior! :)
>
> Scott
>
> -----Original Message-----
> From: nobody@groupstudy.com [mailto:nobody@groupstudy.com] On Behalf Of
> Bryan Bartik
> Sent: Saturday, March 14, 2009 6:21 PM
> To: smorris@internetworkexpert.com
> Cc: naveen M S; GAURAV MADAN; GS CCIE-Lab
> Subject: Re: E1 Vs N OSPF routes
>
> Hmmm....Here is the route entry, metric is 84. Later I will add 64 to make
> it even to the one coming from R4:
>
> R2#sho ip rou 192.168.100.0
> Routing entry for 192.168.100.0/24
> Known via "ospf 1", distance 110, metric 84, type NSSA extern 1
> Last update from 192.168.12.1 on Serial1/1, 00:01:13 ago
> Routing Descriptor Blocks:
> * 192.168.12.1, from 1.1.1.1, 00:01:13 ago, via Serial1/1
> Route metric is 84, traffic share count is 1
>
> This makes the N1 equal to the E1, and E1 is chosen (from ASBR R4)
>
> R2(config)#int s1/1
> R2(config-if)#ip osp cost 128
>
> R2#sho ip route 192.168.100.0
> Routing entry for 192.168.100.0/24
> Known via "ospf 1", distance 110, metric 148, type extern 1
> Last update from 192.168.23.3 on Serial1/0, 00:05:33 ago
> Routing Descriptor Blocks:
> * 192.168.23.3, from 4.4.4.4, 00:05:33 ago, via Serial1/0
> Route metric is 148, traffic share count is 1
>
> We reach your step 3 (or perhaps 2-b-ii)...which shows R2 is picking the
> LSA
> as advertised from R4 (4.4.4.4) instead of R1 (1.1.1.1). R2's costs to both
> routers is 128...(R2-->R1=128, R2-->R4=64+64=128). E1 is chosen over N1
> instead of multiple paths.
>
> The behavior changes if the area between R1 and R2 is a normal area and not
> an NSSA, two routes are installed:
>
> R1(config)#router ospf 1
> R1(config-router)#no are 1 nssa
>
> R2(config)#router osp 1
> R2(config-router)#no are 1 nssa
>
> R2#sho ip rou 192.168.100.0
> Routing entry for 192.168.100.0/24
> Known via "ospf 1", distance 110, metric 148, type extern 1
> Last update from 192.168.23.3 on Serial1/0, 00:01:51 ago
> Routing Descriptor Blocks:
> 192.168.23.3, from 4.4.4.4, 00:01:51 ago, via Serial1/0
> Route metric is 148, traffic share count is 1
> * 192.168.12.1, from 1.1.1.1, 00:01:51 ago, via Serial1/1
> Route metric is 148, traffic share count is 1
>
> So it appears that in a scenario where two equal cost external routes would
> cause multiple paths to be installed, if one of those happened to be a
> type-7 instead, the type-5 would win. I don't find any documentation and I
> hesitate to say it is 100% without testing more. Maybe you could lab it as
> well and share your thoughts.
>
> On Sat, Mar 14, 2009 at 4:17 PM, Scott Morris <
> smorris@internetworkexpert.com> wrote:
>
> > Nope.
> >
> > Re #2, E1s will be preferred over E2's. Then the steps you outline.
> >
> > Re #3, there are still other criteria in here. Type 1's over Type 2's...
> > But if an NSSA route with the P-bit is received, then it will even take
> > precedence over Type 5's. The RFC points this out. I'm not aware of any
> > particular Cisco doc that goes into it.
> >
> > After that, the steps are similar to what you list for #2.
> >
> > Cheers,
> >
> > Scott
> >
> > -----Original Message-----
> > From: nobody@groupstudy.com [mailto:nobody@groupstudy.com] On Behalf Of
> > naveen M S
> > Sent: Saturday, March 14, 2009 5:19 PM
> > To: Bryan Bartik
> > Cc: GAURAV MADAN; GS CCIE-Lab
> > Subject: Re: E1 Vs N OSPF routes
> >
> > Hi Brian,
> >
> > If you still have the setup, can you please paste the "show ip route"
> here
> > ?
> > Wanted to know if router selected R1 or R4 to reach 192.168.100.0. From
> > what
> > I remember about OSPF Path Selection is -
> >
> > 1) OSPF prefers Intra-area path than Inter-area path. "O" route path
> > preferred more than "O-IA" route path.
> >
> > 2) If 2 Type-5 LSAs are received with different ASBRs, then OSPF goes
> > through following path selection rule (per your Cisco Link)
> >
> > a) If metric is different on both LSAs, choose the one with Lowest
> > metric.
> > b) If metric is same on both LSAs, check the Forwarding address in the
> > LSAs.
> > (i) If Forwarding address on both LSAs are set to 0.0.0.0, then
> > choose the nearest ASBR.
> > (ii) If Forwarding address on one is 0.0.0.0 and the other is
> set
> > to non-zero,
> > -> compare metric to ASBR (which set 0.0.0.0) to the metric
> to
> > reach non-zero Fwd'ing address.
> > -> choose the path towards ASBR which has lowest metric.
> > (iii) If Forwarding address is set on both to non-zero values,
> then
> > choose the path towards
> > the forwarding address with the lowest metric.
> >
> > 3) If same external network is learnt as both Type-7 and Type-5 (which is
> > Gaurav's scenario), then ???
> >
> > No cisco documentation is available on this. But very likely it checks
> > for lowest metric first and if they are
> > the same on both LSAs I guess it goes through same selection process
> > described in item (2) above.
> >
> > Rgds,
> > -Naveen.
> >
> >
> > In Gaurav's case, since the external network are learnt as both LSA-7 and
> > LSA-5, I think OSPF would choose lowest metric of both for forwarding
> > traffic.
> >
> > The Cisco link shows the ca
> >
> > On Sat, Mar 14, 2009 at 8:30 AM, Bryan Bartik <bbartik@ipexpert.com>
> > wrote:
> >
> > > Hi Guarav,
> > >
> > > Not sure I understand, but if you had an NSSA, E routes wouldn't be
> > > allowed,
> > > you would only have N routes redistributed by the ASBR. E routes
> wouldn't
> > > make it in, but the N would be converted to E at the ABR...
> > >
> > > Now I guess you could have a scenario where the same route was
> > > redistributed
> > > as a N in one nssa area and E in another regular area and they both
> > showed
> > > up at the NSSA ABR like this:
> > >
> > > R1 and R4 connected to same LAN 192.168.100.0 (no OSPF on it, just
> > > redistributing)
> > > ---R1---R2----R3---R4---
> > >
> > > R2(config-router)#do sho ip osp dat | be pe-7
> > > Type-7 AS External Link States (Area 1)
> > >
> > > Link ID ADV Router Age Seq# Checksum Tag
> > > 192.168.100.0 1.1.1.1 13 0x80000001 0x000E40 0 <----
> > > From R1 in an NSSA
> > >
> > > Type-5 AS External Link States
> > >
> > > Link ID ADV Router Age Seq# Checksum Tag
> > > 192.168.100.0 2.2.2.2 2 0x80000001 0x0084CF 0
> > > <---- Converted to Type 5 by R2
> > > 192.168.100.0 4.4.4.4 14 0x80000001 0x006D55 0
> > <----
> > > From R4 in another area
> > > R2(config-router)#
> > >
> > > In this case, the it looks like the same rules would apply as here:
> > >
> > >
> > >
> >
> >
>
> http://www.cisco.com/en/US/tech/tk365/technologies_tech_note09186a0080124c7d
> >
> .shtml<
> http://www.cisco.com/en/US/tech/tk365/technologies_tech_note09186a008
> 0124c7d%0A.shtml<http://www.cisco.com/en/US/tech/tk365/technologies_tech_note09186a008%0A0124c7d%0A.shtml>
> >
> > >
> > > -hth
> > >
> > > On Sat, Mar 14, 2009 at 8:29 AM, GAURAV MADAN <
> gauravmadan1177@gmail.com
> > > >wrote:
> > >
> > > > Hi
> > > >
> > > > Can someone cearify the rpeference order that ospf follows while
> > > selecting
> > > > between NSSA ; E1 routes (especially when P bit is set ) .
> > > >
> > > > Thnx in advance
> > > > Gaurav Madan
> > > >
> > > >
> > > > Blogs and organic groups at http://www.ccie.net
> > > >
> > > >
> _______________________________________________________________________
> > > > Subscription information may be found at:
> > > > http://www.groupstudy.com/list/CCIELab.html
> > > >
> > > >
> > > >
> > > >
> > > >
> > > >
> > > >
> > > >
> > >
> > >
> > > --
> > > Bryan Bartik
> > > CCIE #23707, CCNP
> > > Sr. Support Engineer - IPexpert, Inc.
> > > URL: http://www.IPexpert.com
> > >
> > >
> > > Blogs and organic groups at http://www.ccie.net
> > >
> > > _______________________________________________________________________
> > > Subscription information may be found at:
> > > http://www.groupstudy.com/list/CCIELab.html
> >
> >
> > Blogs and organic groups at http://www.ccie.net
> >
> > _______________________________________________________________________
> > Subscription information may be found at:
> > http://www.groupstudy.com/list/CCIELab.html
> >
> >
> >
> >
> >
> >
> >
> >
> >
>
>
> --
> Bryan Bartik
> CCIE #23707, CCNP
> Sr. Support Engineer - IPexpert, Inc.
> URL: http://www.IPexpert.com
>
>
> Blogs and organic groups at http://www.ccie.net
>
> _______________________________________________________________________
> Subscription information may be found at:
> http://www.groupstudy.com/list/CCIELab.html
>
>
>
>
>
>
>
>
>
-- Bryan Bartik CCIE #23707, CCNP Sr. Support Engineer - IPexpert, Inc. URL: http://www.IPexpert.comBlogs and organic groups at http://www.ccie.net
This archive was generated by hypermail 2.1.4 : Mon Apr 06 2009 - 06:44:05 ART