IS-IS questions

I'm studying for the bsci, and I'm doing some IS-IS tests. My topology is very simple: there are three routers (ra, rb, and rc), connected via serial lines, each router has a LAN attached.

ra --conn1-- rb --conn2-- rc | | | LANa LANb LANc

LANa: 10.0.0.0/24 LANb: 10.0.1.0/24 LANc: 10.0.2.0/24 conn1: 172.16.0.0/24 (ra is .2, rb is .1) conn2: 172.17.0.0/24 (rb is .1, rc is .2)

IS-IS is enabled on all routers on all interfaces, with the default settings (ie, each router is a l1-l2). All routers are in the same area.

Now, I do a "sh isis database detail" on router ra. The l1 database looks as expected: each router advertises one LSP, containing IP reachability info for the connected IP networks. I expected the l2 database to contain the same info. But instead, the l2 database shows more entries for each LSP:

ra#show isis database detail [cut] IS-IS Level-2 Link State Database: LSPID LSP Seq Num LSP Checksum LSP Holdtime ATT/P/OL ra.00-00 * 0x00000004 0x4234 442 0/0/0 Area Address: 49.0001 NLPID: 0xCC Hostname: ra IP Address: 172.16.0.2 Metric: 10 IS rb.00 Metric: 10 IP 10.0.0.0 255.255.255.0 Metric: 20 IP 10.0.1.0 255.255.255.0 Metric: 30 IP 10.0.2.0 255.255.255.0 Metric: 10 IP 172.16.0.0 255.255.255.0 Metric: 20 IP 172.17.0.0 255.255.255.0 [cut]

why does ra advertise reachability for networks that are not connected (eg, 10.0.2.0 or 172.17.0.0)?

And, how is the address shown in the line "IP Address:" chosen for routers that have more than one IP address? For example, for rb LSP it's 172.17.0.1. Why is that one chosen, instead of 172.16.0.1 or 10.0.1.x? Is this the same "highest IP address" rule used for eg ospf router id?

Thanks for any help.

Reply to
pk
Loading thread data ...

And a last question, why are pseudonode LSP not generated for the attached LANs?

Thanks

Reply to
pk

pk napisa=B3(a):

L1/2 router acts like an ABR in OSPF - it takes all internal routes (L1) =

and represents them in its own level 2 LSP (at least until you use=20 summarization)

Level 1/2 represents its area to the core (level 2), not only its=20 directly connected networks, but all networks from level 1 - look above ;= )

it looks like it does similarly to OSPF (I haven't found it in=20 documentation)

Because these LANs are stub networks - there are no other IS-IS capable=20 routers/neighbors. It would be not efficient to select DIS and create=20 pseudonode on stub networks :)

You should always (when possible) use summarization and set appropriate=20 IS level to every router.

I hope I have explained that :)

Best regards

------------------------ Ern XV535 90' - fajna by=B3a YZF600R Miau :) gg - 1078599; ern ma=B3pa ern waw pl

Reply to
Ern

Yes, but OSPF ABRs advertise networks reachable inside an area *to other areas*, not back to the same area.

However, reading rfc 1195 did prove useful here:

------ The "IP Internal Reachability Information" field may be present in all LSPs transmitted by IP-capable routers. If present, it identifies a list of zero or more [IP address, subnet mask, metrics] reachable by the router which originates the LSP. [cut] If included in level 1 LSPs, this field includes only entries directly reachable by the router which originates the LSP, via one of its interfaces.

->->->-> If included in level 2 LSPs, this field includes only entries reachable by the router which originates the LSP, either via one of its interfaces, or indirectly via level 1 routing.

------

I know that in practice, since internal routers prefer l1 routes, and because one should always set the correct IS level for each router, this is not an issue, but I was curious.

I did some captures with wireshark, and it looks like the address in that line is the one the rfc calls "interface address". What confused me is that, for hello PDUs, the interface address is always the address of the interface out of which the hello is sent (thus different for different neighbors), but for LSPs it is always the same regardless of the neighbor. rfc 1195 is a bit vague here (at least imho):

------ The "IP Interface Address" is included in all IS-IS Hello packets and LSPs transmitted by IP-only and dual routers. In the Hello packets, this field occurs once only, and contains the IP address(es) of the interface on which the Hello packet is transmitted (up to a maximum of 63 IP addresses on each interface). If an IS-IS Hello is transmitted over an interface which does not have an IP address assigned, then this field may be omitted, or may be included with zero entries.

->->->-> In Link State Packets, this field contains a list of one or more IP addresses corresponding to one or more interfaces of the router which originates the LSP. Each IP-capable router must include this field in its LSPs. This field may occur multiple times in an LSP, and may occur in an LSP with any LSP number.

------

And, below:

------ Where a router is both a level 1 and level 2 router, it must include the same IP addresses in its level 1 and level 2 LSPs.

------

It seems that, for LSP PDUs, cisco implementation puts the highest loopback or physical IP address in this field.

Yes, you are right, OSPF does not generate network LSAs for stub networks either. My fault.

Yes, of course. But being a student (vs. one that must manage a production network), I think that, to better understand how things work, one should take the liberty of experimenting a bit and playing around.

Thanks for your time!

Reply to
pk

Yes, but you think in OSPF terms ;) That's because in OSPF area borders=20 fall on routers (ABRs) and in IS-IS on links. So level1/2 IS takes all=20 its level 1 networks and presents them to backbone (level 2) - don't=20 think it resends them to the same area! :)

As you noticed - reading RFCs sometimes helps :)

Regards

--=20

------------------------ Ern XV535 90' - fajna by=B3a YZF600R Miau :) gg - 1078599; ern ma=B3pa ern waw pl

Reply to
Ern

Yes, from the viewpoint of a l2 router talking to another l2 (as, by default, they are all also l2), this makes perfect sense.

Yes, I started reading it just after I posted the original message. I guess next time I'll read it *before* :)

Thanks again!

Reply to
pk

Cabling-Design.com Forums website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.