Shulou(Shulou.com)06/01 Report--

Technorati tags: MPLS,CCIE,Label,LDP, label control

MPLS LDP tag control.

1, mpls ldp advertise-labels [for prefix-access-list [to peer-access-list]]

Label filtering, 1, give 2, receive

By default, the local router assigns labels to all prefixes in the IGP routing table and distributes them to all LDP neighbors.

This command is used to artificially select tags to some LDP or TDP neighbors.

It means, how much to send, whether to send, to whom. All are controlled by the command.

2, mpls ldp neighbor x.x.x.x labels accept ACL_number

This command is that I receive the prefix tags that the ACL sent to me by which neighbor can match.

All other tags are filtered.

Control distribution: to give:

Only basic ACL schemes are supported, prefix lists are not supported.

First of all, the overall situation needs to be typed: no mpls ldp advertise-labels. Just don't distribute the label.

So the above command means:

This device only distributes IGP routing assignment labels of 192.168.254.0 LDP 24 to all IGP neighbors.

Other IGP routing prefixes do not have a label, directly on the directly connected device, except 192.168.254.0According to the label assigned above, the rest are No Labels.

Test: assign label control-send label control

R 3 sent three IGP routing prefix labels to R 2.

As shown in the figure:

Now all you have to do is get R 3 to advertise the label 55.1.1.1 + 32 to R 2. All other labels are filtered out.

Finally, there is a label on R2 for 55.1.1.1Action32, but the other two routing prefixes 33.1.1.1Accord32 and 35.1.1.0Accord24 are No Label states.

ACL solution:

R3:

Mpls label range 300 399

Mpls label protocol ldp

No mpls ldp advertise-labels

Mpls ldp advertise-labels for 2 to 1

!

Interface Loopback0

Ip address 33.1.1.1 255.255.255.255

!

Interface Ethernet0/0

Ip address 35.1.1.3 255.255.255.0

Mpls ip

!

Router ospf 1

Router-id 33.1.1.1

Network 23.1.1.3 0.0.0.0 area 1

Network 33.1.1.1 0.0.0.0 area 1

Network 35.1.1.3 0.0.0.0 area 1

!

Mpls ldp router-id Loopback0 force

Router-id of LDP of access-list 1 permit 22.1.1.1---R2.

Access-list 2 permit 55.1.1.1Murray-tags of prefixes allowed to be released

!

The configuration above, 22.1.1.1 is the ldp router-id of R2.

It means that R3 will assign the label assigned to the routing prefix advertised as 55.1.1.1 to R2Magi 22.1.1.1.

The scheme uses the ACL scheme.

Here's what happens when R2 clears up its ldp neighbors:

There are no tags assigned to 33.1.1.1Compact 32 and 35.1.1.0Universe 24. On the other hand, for 55.1.1.1Compact 32, the label is assigned.

In fact, a tag filter is done on R 3. This is not to say that R3 did not assign labels to 33.1.1.1 and 35.1.1.0 Universe 24, but that no such tags were distributed to R2:

The following is the label binding table for R3:

As you can see here, R3 has tags assigned to 33.1.1.1 and 35.1.1.1.

It just didn't send it to R2.

You can do a lot on the same router.

Mpls ldp advertise's strategy:

Just now it was just an extreme example, because R3 reached the goal of sending only certain tags to R2, but now R5 is miserable, because the configuration of R3 only distributes certain tags to R2. As a result, R5 did not receive any tags from R3:

To solve this problem, continue to release other neighbors' tags on R3:

Configuration of R3:

Mpls label range 300 399

Mpls label protocol ldp

No mpls ldp advertise-labels- does not distribute any tags by default.

Mpls ldp advertise-labels for 2 to 1-advertise labels distributed with the prefixes of 35.1.1.0 Universe 24 and 55.1.1.1 Universe 32 to 22.1.1.1 (R2)

Mpls ldp advertise-labels for 4 to 55 Murray-advertise all prefix tags to 55.1.1.1 (R5)

!

Mpls ldp router-id Loopback0 force

Access-list 1 permit 22.1.1.1

Access-list 2 permit 35.1.1.0

Access-list 2 permit 55.1.1.1

Access-list 4 permit any

Access-list 55 permit 55.1.1.1

!

Finally, on R2, we can see that only 35.1.1.0 and 55.1.1.1 got the tag from R3, while 33.1.1.1 has no label and is No Label:

The following is the label table for R2:

And R5, all the tags are assigned from R3:

Control distribution: label reception

Only basic ACL schemes are supported, prefix lists are not supported.

Now filter the tag reception on R 5.

Under normal circumstances, the label forwarding table on R5 is:

The strategy for making relevant receive tags on R5:

Mpls label range 500 599

Mpls label protocol ldp

Mpls ldp neighbor 33.1.1.1 labels accept 2

!

Mpls ldp router-id Loopback0 force

Access-list 2 permit 12.1.1.0

Access-list 2 permit 22.1.1.1

!

The configuration of R5 above means. I am local to R 5 and only receive labels assigned by neighbor 33.1.1.1 for IGP prefixes 12.1.1.0 Universe 24 and 22.1.1.1 Uniplic32.

You can see here, because there are 12.1.1.0 and 22.1.1.1 tags on R 5 that only receive R 3 advertisements. So only these two prefixes are labeled. All other outgoing label are owned by no Label.

MPLS label Troubleshooting:

As far as tags are concerned, there is nothing to say, because tags are assigned by the system, it is difficult for us to pay attention to control each tag.

However, if the router does not receive the label assigned by the neighbor anyway, the following points must be noted:

1, no ip cef

/ / this is the basis of MPLS. If there is no CEF, no tags will be received. First of all, the forwarding basis of cisco, cef is the core, cef is a collection table extracted from the core routing table, and the LDP tags of MPLS are distributed based on the CEF table prefix, so there is no CEF,LDP and no labels based on the routing table.

Here is a MPLS forwarding table that the router does not have cef turned on:

2, no mpls ip

Cisco system, the global mpls ip is turned on by default and can be turned off manually.

Once turned off, the phenomenon is that show mpls forwarding-table is an empty list item and there is nothing.

And the show mpls ldp binding tag library is also empty. The system will prompt: LIB not enable.

Label information base not eanble. That is, the overall mpls ip has been dried, so we must pay attention to the difference.

If no ip cef, then all tags are No Label. If you kill mpls ip, the label table no longer exists.

(3) the relationship between neighbors in LDP can not come up.

At this time, it is necessary to check whether the transport-IP IGP is mutually reachable.

By default, the router-id of LDP is equal to Transport-IP.

When LDP established neighbors, both routers were sending 224.0.0.2 multicast messages.

Router-id is used to elect who is the initiator of LDP's neighbor establishment.

Transport-IP is used to establish TCP646 neighbors.

Summary: extend the router-id of each other by extending ping at both ends of the device. Pass, and whether the port is blocked or broken, check the IGP routing table.

When troubleshooting, such as:

R1--12.1.1.x--R2

Loop0:11.1.1.1/32 of R1, loop0:22.1.1.1/32 of R2.

On R1 and R2, IGP is the network segment that announces loop0 and 12.1.1.x.

At this time, build another loopback1:11.11.11.11/32 on R1, and then globally type: mpls ldp router-id loop1 force, add the force parameter, the neighbor will down down, but then it will not be established, because R1 does not have the network segment of network loop1 to declare that the loop1 IGP of R2 and R1 is unreachable.

4. Label control [distribute / receive]

Distribution filtering: mpls ldp advertise-labels [for prefix-access-list [to peer-access-list]]

Receive filtering: mpls ldp neighbor x.x.x.x labels accept ACL_number

5. The distribution protocol of the label does not match. TDP & LDP

NOTES: both global and interface can modify the label protocol.

Use the default mpls label protocol ldp on a device. The other device is based on cisco's proprietary TDP protocol.

The two protocols are incompatible. Neighbors cannot be established.

6, the range of the label is too small.

When doing experiments, R1 is often hit with mpls label range 100199.

Because for experiments, 100 tags are almost enough, mainly used to observe phenomena.

In the current network environment, or troubleshooting, if you modify this range very little on a router:

R1 (config) # mpls label range 100 101

The problem is that the downstream router can only receive one prefix with a label, and all the others are No Label.

For example: R1---R2 is directly connected. Do this scope limit on R2:

There is a limit on R 2, and there are only 2 tags available.

So the result of R 1 is a bunch of routing prefixes originating from R 2 without labels.

This is why the existing network does not recommend manual configuration of the scope of the label. Because with the increase of business, the routing prefix is bound to increase, if the restriction is carried out in the early stage, the later fault troubleshooting is easy to make mistakes. In particular, the core equipment, show run are several screens or more than ten screens.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.