IS-IS ATT-bit Control

[Spyridon Katrivesis]

Hi guys,  I am running a lab through dynamips using isis. I have a router R5(L1 router) which is connected to R6(lL1-L2 router). R6 is connected with  other L1-L2 routers. I tried to change the ATT-bit when R6 sends LSP-L1 tou R5 when the backbone area is down but unfortunately is not working.   I tried when the backbone area is down R6 not to advertise LSPs with the ATT bit attached (wich means set to zero(o)) that way no default-route route will appear on the R5 router.  I follow the instructions on the following url; http://www.cisco.com/en/US/products/ps6599/products_white_paper09186a00804fa7a0.shtml " using IS-IS ATT-bit Control Feature"    quick explanation;     I used clns filter-set to match the backbone area which was 49.0002  example: clns filter-set match-49.0002-area permit 49.0002    then I used a route-map to match the clns filter name     router-map is-match-49.0002-area permit 10  match clns address match-49.0002     and then under the router config on R6    router isis .  set-attached-bit route-map is-match-49.0002-area      I am not sure if I am missing anything but when I turn off all the L1-L2 routers except R6 then R6 kept sending LSPs with the att-bit set to 1  and a default route on R5 is pointing to R6 even if the backbone area was down.    Can you please advice?   Many thanks,  Spyros    * I am using IOS version12.2(25)s15

[Giuseppe Larosa]

Hello Spyros,

the IOS image you are using may be not supporting this specific feature.

at the end of the document you have linked:

Cisco IOS Software Release 12.2(4)T

and you have IOS version12.2(25)s15

also if you are going to use ISIS to route IP traffic the ATT is of less importance.

Edit:

looking on the document:

>>Thus, we cannot just use the 39.0000 area address  to satisfy the route map because the Layer 1/Layer 2 routers will still  have the 39.0000 area address in their routing tables. Instead, we can  create another area address in the backbone area as a "beacon" area  address and use it to satisfy the route map

you need to deploy a second CLNS net between two other L2 routers when this net is missing on the ISIS database then the ATT bit will be cleared on L1/L2 routers that check for the presence of the beacon area-id.

The support of multiple nets is used here not for a migration but for providing a CLNS address to be checked.

Hope to help

Giuseppe

[Spyridon Katrivesis]

Hi Giuseppe,  first of all thank you very much for your answer. To be honest I didn't mention that the ATT-bit control feature is not supported it on the IOS that I am currently running.  Unfortunately I am going through my CCIE studies for the SP and I am testing almost everything just to fill confident before I go ahead with anything else.  I am not sure why you say the following and if you can explain further that will be grade;  "also if you are going to use ISIS to route IP traffic the ATT is of less importance."  I think the ATT-bit on the LSPs  is important cause the L1 routers have no other way other than that to route traffic to the other areas on to the other part of the is-is domain.  The L1-L2 router get the responcibility through the ATT bit to add a default route on the L1's routing table. Without that you have to add a static route manually on the L1 router or you have to use route-leaking if you want to learn other routes other than this particular L1 router.  Of cource route-leaking is for resolving suboptimal routing and not for that issue in most cases   If you can explain further that will be grade.   Many thanks, Spyros

[Giuseppe Larosa · Accepted solution]

Hello Spyros,

I guessed it was for your studies, however it can be an interesting feature.

For IP there is an additional tool a default route 0.0.0.0/0

see this taken from our production network:

RT-TO-CRN-P1-E-1#sh isis database | inc 1/0/0
RT-TO-CRN-SM-E.00-00  0x00004CA7   0x34AB        51626             1/0/0  (2)
RT-TO-CRN-SM-E.00-00  0x00003C3B   0xBA80        51626             1/0/0  (10)
RT-TO-CRN-SF-E.00-00  0x0000D013   0x3D06        43249             1/0/0  (18)
RT-TO-CRN-SF-E.00-00  0x0000CFDB   0xE619        47459             1/0/0  (19)
RT-TO-CRN-P1-E-1#

only the first two nodes  are the real exit points to core for IP traffic and are the only ones to inject a default route received from core

sh ip route 0.0.0.0
Routing entry for 0.0.0.0/0, supernet
  Known via "isis", distance 115, metric 100, candidate default path, type level-1
  Redistributing via isis
  Last update from 10.55.130.6 on GigabitEthernet3/47, 7w0d ago
  Routing Descriptor Blocks:
  * 10.55.130.6, from 10.80.34.2, via GigabitEthernet3/47
      Route metric is 100, traffic share count is 1
    10.55.130.2, from 10.80.34.1, via GigabitEthernet1/24
      Route metric is 100, traffic share count is 1

in this sense the ATT bit importance is reduced when routing IP

Hope to help

Giuseppe

[Spyridon Katrivesis]

Hi Guiseppe,

thanks for the extra information. The default route is quit interested.

I assume that the default route is getting the priority because of the admin distance. I hope I am correct.

You will find me asking more questions in the future generally.

I am almost at the end of routing for my studies and I will try to refresh everything and go back to what I left which I couldn't understand.

So you will find me asking more questions please advice if you are familiar with any!

I started the studies and I am walking completely  alone so the only place I can discuss anything is through that forum whcih I am new.

Once more,

Many thanks,

Spyros