Re: Multiarea OSI Routing - General Question

kevin.hart · ‎08-05-2003

I think I've confused myself. Can someone please help with the following? See below

Topology

========

|------NE----NE-----NE------|

| |

NE SDH Ring NE

| Area 100 |

GNE-------------------------GNE

| |

|-GW RouterA-----E1----GW RouterB---|

| |

E1 L2 - Area11 E1

| |

|--Hub RouterA---E1----Hub routerB--|

Hub RouterA 39.840f.8000.0000.0000.0000.0011.0010.0255.0002.00

hub RouterB 39.840f.8000.0000.0000.0000.0011.0010.0255.0001.00

GW RouterA 39.840f.8000.0000.0000.0000.0011.0010.0255.0003.00

GW RouterB 39.840f.8000.0000.0000.0000.0011.0010.0255.0004.00

Hub RouterA has L2 adjacencies with

GW RouterA

Hub RouterB

From Hub RouterA The following OSI Pings work

ping clns 39.840f.8000.0000.0000.0000.0011.0010.0255.0001.00 (Hub RouterB)

Type escape sequence to abort.

Sending 5, 100-byte CLNS Echos with timeout 2 seconds

!!!!!

Success rate is 100 percent (5/5), round-trip min/avg/max = 1/2/4 ms

ping clns 39.840f.8000.0000.0000.0000.0011.0010.0255.0003.00 (GW RouterA)

Type escape sequence to abort.

Sending 5, 100-byte CLNS Echos with timeout 2 seconds

!!!!!

Success rate is 100 percent (5/5), round-trip min/avg/max = 1/3/4 ms

ping clns 39.840f.8000.0000.0000.0000.0011.0010.0255.0004.00 (GW RouterB)

Type escape sequence to abort.

Sending 5, 100-byte CLNS Echos with timeout 2 seconds

CLNS: cannot send ECHO.

Success rate is 0 percent (0/5)

Why does this fail?

Should I be able to ping this router?

donewald · ‎08-05-2003

Question for you. How can you have an L2 adjacency if all your routers are in the same area?

All of your routers in your example are in a single area "39.840f.8000.0000.0000.0000.0011". In this case no L2 adjacencies should be created. Only L1's. You should, if configured correctly have reachability (via L1 adjacencies) from any router to any router. I am unsure by the NETs you've provided where your Area 100 come into play unless this is part of your Multiarea on your GW router A and B. If this is the case then a L2 would be required for GW A and B to have reachability and advertise the Area 100 within their L2 LSPs.

Hope this helps,

Don

kevin.hart · ‎08-05-2003

Thanks for your quick response.

Please correct me if I'm wrong. L2 routers form L2 adjacenies with L2 or L1/L2 neighbors if in the same area or between areas. see show clns nei from "Hub RouterA" router

sh clns nei

Area L2BackBone:

System Id Interface SNPA State Holdtime Type Protocol

lab-GW-A-dcn-r1Se0/0 *HDLC* Up 24 L2 IS-IS

lab-LSH-dcn-r1 Se0/1 *HDLC* Up 23 L2 IS-IS

I can only ping the L2 routers that I have a direct adjacency with.

Yes, you are correct AREA 100 is part of my multiarea config and sites on GW A and GW B. Sorry, I should have not included this on the diagram

donewald · ‎08-05-2003

L2 routers should only form adjacencies with routers in different areas. In the same area (NET)L1's should be formed. Every isis interface is default L1/L2 so depending on your configuration you could potentially have both on a single interface. A "show clns route" should tell you if you are learning other areas (which I did not see included in your example). I would assume that your routers not connected to Area 100 would not have this route due to the L1 issue I mentioned. Might be why you are having reachability issues.

Hope this helps,

Don

kevin.hart · ‎08-05-2003

"L2 routers should only form adjacencies with routers in differenct areas". Sorry I would disagree with this. see below

"show clns nei" - from Hub RouterA - This shows that we have formed a L2 adjacency.

lab-LSH-dcn-r1 is Hub RouterB

sh clns nei

Area L2BackBone:

System Id Interface SNPA State Holdtime Type Protocol

lab-GW-A-dcn-r1Se0/0 *HDLC* Up 26 L2 IS-IS

lab-LSH-dcn-r1 Se0/1 *HDLC* Up 23 L2 IS-IS

"show clns route" from Hub RouterA

sh clns route

Codes: C - connected, S - static, d - DecnetIV

I - ISO-IGRP, i - IS-IS, e - ES-IS

C 39.840f.8000.0000.0000.0000.0002 [2/0], Local IS-IS Area

C 39.840f.8000.0000.0000.0000.0011 [2/0], Local IS-IS Area

C 39.840f.8000.0000.0000.0000.0002.0010.0255.0002.00 [1/0], Local IS-IS NET

C 39.840f.8000.0000.0000.0000.0011.0010.0255.0002.00 [1/0], Local IS-IS NET

i 39.840f.8000.0000.0000.0000.0001 [110/15]

via lab-LSH-dcn-r1, Serial0/1

i 39.840f.8000.0000.0000.0000.0100 [110/10]

via lab-GW-A-dcn-r1, Serial0/0

!

"ISIS config from Hub RouterA"

router isis L2BackBone

net 39.840f.8000.0000.0000.0000.0011.0010.0255.0002.00

redistribute connected metric 10 route-map CONNECTED-TO-ISIS

redistribute static ip route-map STATIC-TO-ISIS

passive-interface Loopback0

maximum-paths 6

is-type level-2-only

log-adjacency-changes

!

"ISIS config from Hub Router B"

!

router isis L2BackBone

net 39.840f.8000.0000.0000.0000.0011.0010.0255.0001.00

redistribute connected metric 10 route-map CONNECTED-TO-ISIS

redistribute static ip route-map STATIC-TO-ISIS

passive-interface Loopback0

maximum-paths 6

is-type level-2-only

log-adjacency-changes

!

"ISIS config from GW A"

!

router isis L2BackBone

net 39.840f.8000.0000.0000.0000.0011.0010.0255.0003.00

redistribute connected metric 10 route-map CONNECTED-TO-ISIS

redistribute static ip route-map STATIC-TO-ISIS

passive-interface Loopback0

maximum-paths 6

is-type level-2-only

log-adjacency-changes

!

router isis AREA100

net 39.840f.8000.0000.0000.0000.0100.0010.0255.0003.00

is-type level-1

partition avoidance

!

"ISIS config from GW B"

!

router isis L2BackBone

net 39.840f.8000.0000.0000.0000.0011.0010.0255.0004.00

redistribute connected

is-type level-2-only

log-adjacency-changes

!

router isis AREA100

net 39.840f.8000.0000.0000.0000.0100.0010.0255.0004.00

is-type level-1

partition avoidance

!

donewald · ‎08-05-2003

You are correct, my mistake. L2 routers can be in a contiguous area all having adjacencies and all being L2. I'll poke around a bit in your previous posts to see if I can spot the issue.

Regards,

Don

ruwhite · ‎08-05-2003

This is odd--are you certain all of your routers have the right net addresses? In your original explanation, you say that GA A should be 39.840f.8000.0000.0000.0000.0011.0010.0255.0003.00, but in the actual routing information above, I see 39.840f.8000.0000.0000.0000.0100, which looks transposed? Or is this from your "other" routing area?

Aren't these two "routing areas" actually in the same domain/area? How are you seperating the routing information in this case?

Russ.W

kevin.hart · ‎08-06-2003

Hi

39.840f.8000.0000.0000.0000.0100 is from my "other" routing area. This is a L1-only area. Each GW router will terminate multiple L1 areas.

I’ll try to explain the set-up in a little more detail.

The topology is a square with routers sitting in each corner. The routers are then connected by E1’s. Each router is L2-only and routing for area 39.840f.8000.0000.0000.0000.0011.

GW-A--------------GW-B

| |

CRC-dcn-r1-----LSH-dcn-r1

The “show isis database l2” is taken from lab-CRC-dcn-r1

The two gateway routers lab-GW-A-dcn-r1 & lab-GW-B-dcn-r1 are used to terminate each L1 area.

See “show isis database l2” from lab-CRC-dcn-r1

sh isis database l2

Area L2BackBone:

IS-IS Level-2 Link State Database:

LSPID LSP Seq Num LSP Checksum LSP Holdtime ATT/P/OL

lab-LSH-dcn-r1.00-00 0x000021E0 0x8D4E 786 0/0/0

lab-CRC-dcn-r1.00-00* 0x000009E4 0xC83F 1022 0/0/0

lab-GW-A-dcn-r.00-00 0x00000A05 0x2DDD 1004 0/0/0

lab-GW-B-dcn-r.00-00 0x00000080 0xFF77 365 0/0/0

One other thing to point out is that I’m running in a dual environment (IP and CLNS). IP is fine i.e. from lab-CRC-dcn-r1 I can ping the Loopback address for lab-GW-B-dcn-r1. See below

lab-CRC-dcn-r1>ping 10.255.0.4

Type escape sequence to abort.

Sending 5, 100-byte ICMP Echos to 10.255.0.4, timeout is 2 seconds:

!!!!!

Success rate is 100 percent (5/5), round-trip min/avg/max = 4/4/4 ms

If I ping the NET for lab-GW-B-dcn-r1 it fails.

lab-CRC-dcn-r1#ping clns 39.840f.8000.0000.0000.0000.0011.0010.0255.0004.00

Type escape sequence to abort.

Sending 5, 100-byte CLNS Echos with timeout 2 seconds

CLNS: cannot send ECHO.

Success rate is 0 percent (0/5)

lab-CRC-dcn-r1#

One last thing,

Doing a “show clns route” from lab-CRC-dcn-r1 shows I’m learning the following routes.

lab-CRC-dcn-r1#sh clns route

Codes: C - connected, S - static, d - DecnetIV

I - ISO-IGRP, i - IS-IS, e - ES-IS

C 39.840f.8000.0000.0000.0000.0002 [2/0], Local IS-IS Area

C 39.840f.8000.0000.0000.0000.0011 [2/0], Local IS-IS Area

C 39.840f.8000.0000.0000.0000.0002.0010.0255.0002.00 [1/0], Local IS-IS NET

C 39.840f.8000.0000.0000.0000.0011.0010.0255.0002.00 [1/0], Local IS-IS NET

i 39.840f.8000.0000.0000.0000.0001 [110/15]

via lab-LSH-dcn-r1, Serial0/1

i 39.840f.8000.0000.0000.0000.0100 [110/20]

via lab-GW-A-dcn-r1, Serial0/0

If I disable routing for “39.840f.8000.0000.0000.0000.0100” from lab-GW-A-dcn-r1.

It now routes via lab-LSH-dcn-r1

lab-CRC-dcn-r1#sh clns route

Codes: C - connected, S - static, d - DecnetIV

I - ISO-IGRP, i - IS-IS, e - ES-IS

C 39.840f.8000.0000.0000.0000.0002 [2/0], Local IS-IS Area

C 39.840f.8000.0000.0000.0000.0011 [2/0], Local IS-IS Area

C 39.840f.8000.0000.0000.0000.0002.0010.0255.0002.00 [1/0], Local IS-IS NET

C 39.840f.8000.0000.0000.0000.0011.0010.0255.0002.00 [1/0], Local IS-IS NET

i 39.840f.8000.0000.0000.0000.0001 [110/15]

via lab-LSH-dcn-r1, Serial0/1

i 39.840f.8000.0000.0000.0000.0100 [110/25]

via lab-LSH-dcn-r1, Serial0/1

Hope this helps

Kevin

kevin.hart · ‎08-06-2003

All

I've been playing in the lab.

I've changed my L2 routers to be L1 routers and I now have full connectivity. i.e. I can ping all net's.

lab-LSH-dcn-r1#ping 39.840f.8000.0000.0000.0000.0011.0010.0255.0001.00

Translating "39.840f.8000.0000.0000.0000.0011.0010.0255.0001.00"

Type escape sequence to abort.

Sending 5, 100-byte CLNS Echos with timeout 2 seconds

!!!!!

Success rate is 100 percent (5/5), round-trip min/avg/max = 1/1/4 ms

lab-LSH-dcn-r1#ping 39.840f.8000.0000.0000.0000.0011.0010.0255.0002.00

Translating "39.840f.8000.0000.0000.0000.0011.0010.0255.0002.00"

Type escape sequence to abort.

Sending 5, 100-byte CLNS Echos with timeout 2 seconds

!!!!!

Success rate is 100 percent (5/5), round-trip min/avg/max = 1/2/4 ms

lab-LSH-dcn-r1#ping 39.840f.8000.0000.0000.0000.0011.0010.0255.0003.00

Translating "39.840f.8000.0000.0000.0000.0011.0010.0255.0003.00"

Type escape sequence to abort.

Sending 5, 100-byte CLNS Echos with timeout 2 seconds

!!!!!

Success rate is 100 percent (5/5), round-trip min/avg/max = 4/4/8 ms

lab-LSH-dcn-r1#ping 39.840f.8000.0000.0000.0000.0011.0010.0255.0004.00

Translating "39.840f.8000.0000.0000.0000.0011.0010.0255.0004.00"

Type escape sequence to abort.

Sending 5, 100-byte CLNS Echos with timeout 2 seconds

!!!!!

Success rate is 100 percent (5/5), round-trip min/avg/max = 1/2/4 ms

lab-LSH-dcn-r1#

Question is why don't I have full connectivity when they are L2?

ruwhite · ‎08-06-2003

So, two things.... First, I think you might be confusing the router by having the same net addresses under both processes. Only one process, I think, is going to be able to install a route in the CLNS routing table, so one is probably bumping the other out (?). While the two processes don't share the same database, they do share the same routing table.

That may actually account for why it works with l1 routing, and not l2 routing--it may not be working in the _other_ area when one of the two is configured as an l1, and the other is configured as an l2.

I'm also suspicious that, for some reason, your l2 area may not be contiguous (?). I actually don't know why it should matter, since all of the actual routes are in the same domain/area, so we shouldn't be looking for an l2 border anyway, but it just seems odd that I haven't seen where hub a knows anything abotu the .004.

Russ.W

kevin.hart · ‎08-06-2003

Sorry, I might have confused you. I don't have the same net address configured under both processes. See below for config from .004 router

!

router isis L2BackBone

net 39.840f.8000.0000.0000.0000.0011.0010.0255.0004.00

redistribute connected metric 10 route-map CONNECTED-TO-ISIS

is-type level-2-only

log-adjacency-changes

!

router isis AREA100

net 39.840f.8000.0000.0000.0000.0100.0010.0255.0004.00

is-type level-1

!

ruwhite · ‎08-06-2003

Ah, nevermind, I see now--they are different. Okay, so I'm not certain why it would do this with level 2 routing, and not level 1. I suspect the level 2 domain may not be contiguous.... I wonder.... Where does clns ping source from when you ping? What interface/net id? I wonder if it's sourcing from the other process' net for some reason, and the destination doesn't have a route back, since they are both L2, and it may just pick one. When you're pinging an L1 only address, it might be able to look in the database, and figure out where to source the packet from, so the destination has a path back (?).

It doesn't look that way--it looks like the source router is saying it can't reach that destination, not that the destination router isn't receiving the pings. We should be preferring the L1 route anyway, the L2 routes shouldn't even be in the table here. But you could find out with a debug clns packet on the destination router, probably, to see if you're getting the pings, and just not responding.

Could you look at the table for the specific route to the destination you're trying to ping, as well? I don't remember seeing it in there from your last postings....

I might have to set this up in my lab, and see if I can replicate it to figure out what's going on. It might just be best to send me your configs in email, so I can try and put it up in my lab here, and figure out what's going on.

Russ.W