EIGRP Query Range and summarisation

geezer

Hi

Would someone clarify why queries aren't forwarded by neighbour routers when a summarised route is advertised by that router to limit query range and prevent SIA occurences? What I have read is that the router receiving the query responds with "network unreachable" message to the summarising router.

I thought that if a router doesn't know of a network it forwards queries out of I/F's apart from the one it received the original query from.

Any clarification appreciated.

networker050184

The summaries are used to summarize what is behind that router. If the route isn't included in the summary, which means the route is not behind this router, there is no point in the queries going any further.

geezer

I understand that traffic is forwarded to null0 but the self-study guide I am using on page 125 (Dianne Teare) states that the summarising router sends a more detailed route to the router which received the summarised route. This doesn't make sense as surely the summarising router wouldn't forward the route anyhow?

APA

I think you've mixed up a few terms....

If a router receives a query it must respond to that query....... This is why SIA queries are there as well to ensure links stay up between two routers that have no reason to tear down their neighbour adjacency because a router further down the link is waiting for a response from it's neighbouring router.

A query is sent as a reliable pkt (remember the Reliable Tranport Protocol is used by EIGRP to send certain pkts as reliable packets to ensure guranteed delivery.... Which means the receiving router must respond!!! So if it receives a query on an interface it must respond back out that interface to the neighbour the query was sent from)

When it comes time to receiving and forwarding routes the router will not forward a route back out an interface that the route was received on. When traffic comes through to a router that has advertised a summary route but it does not have a explicit entry in it's route table for this traffic then it is forwarded to null0 to ensure a routing loop doesn't form.

Hope this helps

geezer

What I don't understand though, is why a router that is summarising a route of say 172.30.5.0/24 that is querying 172.30.0.0/16 forwards the packet when the downstream router will be informed of the summarised route only.

Probably too tired at present (very late ) but all I can think of is that the router will look for the exact route match which is why the summarising router forwards it. I can see how the summarising router might forward the query about the 172.30.5.0/24 as it will have an exact match in its routing table, BUT it also summarises the route downstream too though?

Is the query forwarded until an infinite route or valid route reply is received regardless of summarisation which limits how many hops the query has to travel?

gojericho0

geezer wrote:

What I don't understand though, is why a router that is summarizing a route of say 172.30.5.0/24 that is querying 172.30.0.0/16 forwards the packet when the downstream router will be informed of the summarized route only.

It will still do a query even for the specific route even though it falls under the summary is a function of DUAL. The only way to prevent this is to configure the neighboring interface as a stub.

Same thing happens if a router loses its Ethernet connection. It will still send queries out serial interfaces even there is no other path. That is why i truly consider EIGRP a distance vector router and not really a hybrid. It still routes by rumor and does not have the big picture of a link-state protocol

geezer

The summarising router then has to query a neighbour as IT has the exact route in ITS routing table. It is up to the neighbour with only the summary route to mark the more specific route as unreachable and inform the sending router of this.

The router sees the network on the ethernet port as just another network so has to send updates out other interfaces in case there is a successor route for it. EIGRP is an advanced distance vector routing protocol and believes what it's neighbours tell it. The function of DUAL keeps routing updates to a minimum unlike RIP and IGRP and doesn't flood updates to all routers like OSPF does when a network change happens though I am a bit rusty on that and that is my next 2 chapters - yippee!!

Thanks, back to fun reading.

singh8281

gojericho0 wrote:

[Same thing happens if a router loses its Ethernet connection. It will still send queries out serial interfaces even there is no other path. That is why i truly consider EIGRP a distance vector router and not really a hybrid. It still routes by rumor and does not have the big picture of a link-state protocol

I haven't lab this scenario up but what if the connected loopback goes down, will queries be sent out to neighboring routers or will the router just poison the route and inform the neighbors??

gojericho0

If EIGRP is running on the loopback interface, EIGRP will query

marco71

singh8281 wrote:

I haven't lab this scenario up but what if the connected loopback goes down, ... ??

A loopback interface, once declared, never goes down

Inc

marco71 wrote:

singh8281 wrote:

I haven't lab this scenario up but what if the connected loopback goes down, ... ??

A loopback interface, once declared, never goes down

Interface IP-Address OK? Method Status Protocol

Loopback0 unassigned YES manual administratively down down

singh8281

marco71 wrote:

singh8281 wrote:

I haven't lab this scenario up but what if the connected loopback goes down, ... ??

A loopback interface, once declared, never goes down

True, but i was talking about manually shutting it down.

kpjungle

Hi,

I was wondering in addition to this, if there is there a scenario in which a downstream router, would simply never have had an exact match to a network/route (say. 172.16.1.0/24) and if so, would that router further propagate the query?

I cant really come up with any scenario in which any router in a network wont know of all routes (or part of them because of summarization, either auto or manually), so Im not sure if the above is even a valid question in that regard.

APA

kpjungle wrote:

Hi,

I was wondering in addition to this, if there is there a scenario in which a downstream router, would simply never have had an exact match to a network/route (say. 172.16.1.0/24) and if so, would that router further propagate the query?

I cant really come up with any scenario in which any router in a network wont know of all routes (or part of them because of summarization, either auto or manually), so Im not sure if the above is even a valid question in that regard.

If a normal EIGRP router and it had another downstream router connected to it then yeah it would propagate the query, if it was the only downstream router then it would have no need to propagate the query and just respond back saying it has no route to that network.

Now if it was a EIGRP stub then it would automatically respond back saying inaccessible as stub routers only advertise what they have (if configured to advertise) or are set to receive-only mode where if queries are received they are pretty much told to buzz off....

singh8281

kpjungle wrote:

I was wondering in addition to this, if there is there a scenario in which a downstream router, would simply never have had an exact match to a network/route (say. 172.16.1.0/24) and if so, would that router further propagate the query?

If the neighbor router doesn't have a route in its routing table or has a summary route and route being queried is part of the summary route, the router will not forward query further, it will stop the query from propagating further and send a network not found message back to the querier.

telco1

kpjungle wrote:

Hi,

I was wondering in addition to this, if there is there a scenario in which a downstream router, would simply never have had an exact match to a network/route (say. 172.16.1.0/24) and if so, would that router further propagate the query?

I cant really come up with any scenario in which any router in a network wont know of all routes (or part of them because of summarization, either auto or manually), so Im not sure if the above is even a valid question in that regard.

everything is locally significant in a way. the ASBR will propagate a default route down to all the other routers and they will always no how to get out. if for some reason a device is unable to reach a network, due to a down router, or broken link, the router will begin the query/reply sequence until it finds a path. if it is unable to find a path, the route is placed in as a SIA, and the router will continue to query. the querys are sent out as multicasts to all the other devices.