[j-nsp] Juniper BGP Route Metrics

Otto Kreiter otto.kreiter at dante.org.uk
Wed May 16 11:48:45 EDT 2007


Hi Dan,

Is LDP enabled in your whole network or just on some of  the routers ?

I expect that all your BGP next-hops, or at least the ones from the LDP
enabled routers,  to be present in the inet.3 route table, and because
BGP inspects inet.3 as well for BGP next-hops and LDP routes have a
lower preference that OSPF the LDP route will be selected to forward the
packets.

You can check it with:
show route table inet.3

One  option may be to raise the preference of the LDP routes and then
your IGP routes in inet.0 may have a better preference or disable
"traffic-engineering bgp" which if I remember correctly is enabled by
default and enables BGP to use the routesin inet.3. But I'm not sure
which will suit you.

I hope it helps.
Cheers,
Otto

Dan Benson wrote:
> Alan and all, I am currently tracking IGP metrics in LDP.  Please see my 
> show route proto bgp detail below.  I have changed my router-id's just 
> to stop and punks form trying to have fun with all the spare time they 
> have..  Hope you understand.  Thanks again for the help.. //db
>
> show route protocol bgp 155.212.0.0 detail (Random subnet learned from a 
> peer AS):
>
> inet.0: 217037 destinations, 706742 routes (217035 active, 2 holddown, 0 
> hidden)
> 155.212.0.0/16 (3 entries, 1 announced)
>        *BGP    Preference: 170/-91
>                Next-hop reference count: 35478
>                Source: 10.0.0.16
>                Next hop: via at-0/0/0.0, selected
>                Label operation: Push 198480
>                Protocol next hop: 198.32.118.12
>                Indirect next hop: 93403a8 262226
>                State: <Active Int Ext>
>                Local AS:  1784 Peer AS:  1784
>                Age: 4d 14:15:14        Metric: 0       Metric2: 0
>                Task: BGP_1784.10.0.0.16+2131
>                Announcement bits (3): 0-KRT 5-RT 6-Resolve tree 2
>                AS path: 174 14751 14751 14751 I ()
>                Communities: 1784:1001
>                Localpref: 90
>                Router ID: 10.0.0.16
>         BGP    Preference: 170/-91
>                Next-hop reference count: 11931
>                Source: 10.0.0.80
>                Next hop: via at-0/0/0.0, selected
>                Label operation: Push 198576
>                Protocol next hop: 198.32.124.103
>                Indirect next hop: 12734750 262165
>                State: <Int Ext>
>                Inactive reason: Router ID
>                Local AS:  1784 Peer AS:  1784
>                Age: 4d 14:15:31        Metric: 0       Metric2: 0
>                Task: BGP_1784.10.0.0.80+179
>                AS path: 174 14751 14751 14751 I ()
>                Communities: 1784:1001
>                Localpref: 90
>                Router ID: 10.0.0.80
>         BGP    Preference: 170/-91
>                Next-hop reference count: 11682
>                Source: 10.0.0.113
>                Next hop: via at-0/0/0.0, selected
>                Protocol next hop: 198.32.176.131
>                Indirect next hop: 9340444 262289
>                State: <Int Ext>
>                Inactive reason: Router ID
>                Local AS:  1784 Peer AS:  1784
>                Age: 4d 14:14:59        Metric: 0       Metric2: 0
>                Task: BGP_1784.10.0.0.113+3209
>                AS path: 174 14751 14751 14751 I ()
>                Communities: 1784:1001
>                Localpref: 90
>                Router ID: 10.0.0113
>
> As you can see, I am not getting ANY metrics or metrics2 in my routes 
> learned from internal peers.  I would like to see a metric that is 
> directly associated with the IGP metric from the local router to the 
> edge router's loopback but I am starting to think it my just be a pipe 
> dream.   Thanks again.. //db
>
> Alan Gravett wrote:
>   
>> Dan,
>>  
>> I suggest you double check whether the routers are configured as 
>> suggested -
>> [ edit protocols ldp ]
>> track-igp-metric
>>  
>> If the route selection is based on BGP as you suggest and you are 
>> running LDP everywhere then
>> there should not be any change to the BGP next-hop selected, suspect 
>> the tie-breaker is
>> IGP metric. The default metric for LDP is 1, which is why you need to 
>> "track-igp-metric".
>>  
>> Alan
>>
>>  
>> On 5/16/07, *Dan Benson* <dbenson at swingpad.com 
>> <mailto:dbenson at swingpad.com>> wrote:
>>
>>     Piotr, Thank you for the thought.  I am currently happy with the LDP
>>     aspect of the network and yes I am tracking IGP for LDP learned
>>     routes but
>>     my issue is purely in BGP and how the routes go all the way down
>>     the line
>>     to the tie breaker of router-id.  Any others thoughts are
>>     welcome.  Thanks
>>     again.. //db
>>
>>     On Tue, May 15, 2007 5:00 pm, Piotr Marecki wrote:
>>     > Hi Dan,
>>     >
>>     >
>>     > Are you using LDP ? Try "track-igp-metric" under [protocols ldp]
>>     stanza.
>>     >
>>     >
>>     > regards
>>     >
>>     > Piotr Marecki
>>     >
>>     >
>>     > ----- Original Message -----
>>     > From: "Dan Benson" <dbenson at swingpad.com
>>     <mailto:dbenson at swingpad.com>>
>>     > To: <juniper-nsp at puck.nether.net
>>     <mailto:juniper-nsp at puck.nether.net>>
>>     > Sent: Tuesday, May 15, 2007 10:08 PM
>>     > Subject: [j-nsp] Juniper BGP Route Metrics
>>     >
>>     >
>>     >
>>     >> All, I am in a funny spot in my network where I could you a little
>>     >> assistance.  I run a fairly widespread network in the US, where
>>     I peer
>>     >> and buy upstream bandwidth in multiple locations in may cases
>>     with the
>>     >> same Networks.
>>     >>
>>     >> We currently run OSPF as in IGP and carry IBGP over loopbacks
>>     for bgp
>>     >> route distribution. Recently we have implemented MPLS for many
>>     obvious
>>     >> reasons and in doing so changed the way our customer traffic flows
>>     >> dramatically.  This from what I can tell is due to route
>>     preference,
>>     >> where before MPLS a flow would be looked up at every hop and
>>     might be
>>     >> handed to a connected AS more locally to the origination of the
>>     flow (A
>>     >>  good thing).  Now with MPLS when the flow is setup it is never
>>     looked
>>     >> at again from an IP standpoint and therefore leaves my route
>>     preference
>>     >> using the last resort, router-id.  This as anyone knows is
>>     plain old
>>     >> wrong in this case and is hurting my traffic in more ways then one.
>>     >>
>>     >> Looking and testing I have had absolutely no luck setting the
>>     IBGP core
>>     >>  to use IGP metrics for route preference, nor have I had any
>>     luck with
>>     >> external router id (At this point I will try anything).  My
>>     OSPF links
>>     >> are all set with specific Metrics and flow just the way they
>>     should but
>>     >>  my hope is that I can get my BGP route table to use either the
>>     metric
>>     >> or the metric2 as the last resort tie breaker for prefix next-hops
>>     >> instead of the IBGP router-id.
>>     >>
>>     >> I have tested this in junos 7.3 and 8.2 with no luck.  I have also
>>     >> tried policies such as metric add 0 but still have no metrics
>>     in the bgp
>>     >> route table.  Have I been assuming that this feature does
>>     something
>>     >> completely different then what I have read it can do?  Does
>>     anyone out
>>     >> there have any suggestions as to how to use the most locally
>>     (Closest
>>     >> OSPF metric'd
>>     >> IBGP neighbor) learned next-hop for a prefix?  Am I left to
>>     manually set
>>     >>  all the metrics on outbound routes in my core to force each
>>     router to
>>     >> use the more locally learned route?
>>     >>
>>     >> I thank you all for your help in advance and take care.. //db
>>     >>
>>     >>
>>     >>
>>     >> _______________________________________________
>>     >> juniper-nsp mailing list juniper-nsp at puck.nether.net
>>     <mailto:juniper-nsp at puck.nether.net>
>>     >> https://puck.nether.net/mailman/listinfo/juniper-nsp
>>     >>
>>     >
>>
>>
>>     _______________________________________________
>>     juniper-nsp mailing list juniper-nsp at puck.nether.net
>>     <mailto:juniper-nsp at puck.nether.net>
>>     https://puck.nether.net/mailman/listinfo/juniper-nsp
>>
>>
>> ------------------------------------------------------------------------
>>
>> No virus found in this incoming message.
>> Checked by AVG Free Edition. 
>> Version: 7.5.467 / Virus Database: 269.7.1/805 - Release Date: 5/15/2007 10:47 AM
>>   
>>     
>
> _______________________________________________
> juniper-nsp mailing list juniper-nsp at puck.nether.net
> https://puck.nether.net/mailman/listinfo/juniper-nsp
>   



More information about the juniper-nsp mailing list