reditribution&prefix-list

[Antonio_1_2]

Hello,

I have this configuration:

router eigrp 1

autonomous-system 1

network 10.10.0.0 0.0.0.3

no auto-summary

redistribute static route-map MAP1

redistribute bgp 65001 route-map MAP1

exit-address-family

!

ip prefix-list NETS seq 10 permit 10.0.0.0/24 le 32

ip prefix-list NETS seq 20 permit 20.0.0.0/24 le 32

ip prefix-list NETS seq 30 permit 30.0.0.0/24 le 32

!

route-map MAP1 permit 10

match ip address prefix-list NETS

set metric 1000000 1 255 1 1500

When I remove from prefix list ip prefix list NETS permit 20.0.0.0/24 le 32, route 20.0.0.0/24 is stil redistributed and advertised (and I can see it with show ip eigrp topology 20.0.0.0/24) until I issue commad clear ip route *.

Is it always necessary to issue command clear ip route * when something is changed in prefix-list (regarding redistribution)? And does this apply for all dynamic routing protocols (EIGRP, OSPF, BGP, RIP) when redistribution is used?

Thanks in advance,

A.

[milan.kulik · Accepted solution]

Hi Antonio,

I'd say YES, this is the best practise.

Even while some routing protocols are sending the whole route table periodically (RIP, OSPF), if you change the prefix-list or distribution filter the receiving protocol might never receive the updated info for specific prefixes and they might stay in the routing table for ever theoretically.

BR,

Milan

[Giuseppe Larosa]

Hello Milan,

I think redistribution has its own timers and that that should soon or later converge.

Using the clear ip route * speeds up the process and it is good practice.

Hope to help

Giuseppe

[milan.kulik]

Hi Giuseppe,

I'd like to believe so.

But never seen any such a timer documented.

I was told by a friend of mine there was some BGP timer checking the BGP table for "never updated" prefixes, e.g.

But again, nothing documented, I'm afraid.

BR,

Milan

[Giuseppe Larosa]

Hello Milan,

I've noticed that ACLs used in route-map have matches that increment over time.

My understanding is that this is caused by the re-execution of some IOS code related to redistribution.

I may be wrong but this is my idea otherwise why should these counters increment without an explicit action like clear ip route x.x.x.x y.y.y.y ?

Hope to help

Giuseppe

[milan.kulik]

Hi Giuseppe,

well, if something changes or complete routing table (OSPF, e.g.) is sent, then the ACL counters will increase.

I might be wrong, of course, but do they increase also for redistributing static, e.g., if nothing changes?

My fear was that if the modified prefix-list would prevent sending updated info about a specific subnet (20.0.0.0/24 in the original case), the routing protocol TO which we are redistributing will never remove the subnet from the table.

BR,

Milan

[Antonio_1_2]

Thanks Milan,

So theoreticaly if I used OSPF (or RIP), change would be noticed after 30 minutes (period when OSPF sends whole routing table) and in my case route 20.0.0.0/24 would disapear without issuing clear ip route *?

regards,

Vedran

[milan.kulik]

Hi Vedran,

this is what I'm not sure.

As you are removing the line from your prefix-list, my understanding is your OSPF will not send any 20.0.0.0/24 info through the new prefix list.

So your EIGRP will not receive any info about the subnet even within the full OSPF routing table update.

I'm not sure what EIGRP will do with the old 20.0.0.0/24 subnet info - hopefully it will remove it.

I'm afraid the only chance is a test and observe.

BR,

Milan

[milan.kulik]

And if not removed,

clear ip route 20.0.0.0 255.255.255.0

should remove it without disrupting any other routing.

BR,

Milan