Policy Based Routing

Policy-Based routing is a neat trick to tweak traffic streams.

For Policy Based Routing to work you will the following;
– Define traffic stream in an access-list.
– Write down a policy.
– Determine the inbound interface.

First use an ACL to define the traffic streams.
You can use standard, extended or named access-list.

In this example traffic sourced from and destined for is defined in ACL 100.

access-list 100 permit ip host host
access-list 100 permit ip host host

Next you will to define the policy.
If you want to want set the next hop for this traffic stream. You have to do;

route-map just-a-name permit 10
 match ip address 100
 set ip default next-hop

Once you have formulated the traffic stream and the tweak all you have to do is apply it to an interface.

interface Ethernet0/0
 ip address
 ip policy route-map just-a-name

Be careful : the name of the route-map is CaSe senSitive.

Why is this a great little trick; you can use policy based routing to prevent asymetric routing in a HSRP setup. Or you can send specific traffic towards an IDS / firewall /deep packet inspection device. Or simply send traffic to an black-hole in case of unwanted/malicious traffic.


Posted in CCNP | Tagged ,


When a network port encounters an error. This port is automatically shutdown to ensure network stability. When error-disable is enabled.

The predefined error are as follows;


Primary function of error-disable is fault detection. Secondary you can set up recovery, per predefined error.  This can be done by ;

errdisable recovery cause

Be sure to set the recovery interval to your specific needs;

errdisable recovery interval 60

Whenever  a port is disabled due to an error-disable condition you should see something like ;

somerouter#show interface GigabitEthernet1/6
GigabitEthernet1/6 is down, line protocol is down (err-disabled)
--//output omitted for brevity//--

Interesting to see that the port is in down/down status with an err-disabled condition.

If you do not resolve the root cause the log will be flooded with

Jan 10 08:51:35.074 CET: %PM-4-ERR_RECOVER: Attempting to recover from link-flap err-disable state on Gi1/6
Jan 10 08:51:37.179 CETT: %PM-4-ERR_DISABLE: link-flap error detected on Gi1/6, putting Gi1/6 in err-disable state

To resolve the problme one has to know the root cause, for this you’ll have to issue the command “show errdisable recovery”. First you see a list of errors where recovery is enabled, or disabled. (See column ‘Timer Status’)

somerouter#sh errdisable recovery
ErrDisable Reason            Timer Status
-----------------            --------------
arp-inspection               Disabled
bpduguard                    Enabled
channel-misconfig            Enabled
dhcp-rate-limit              Disabled
gbic-invalid                 Enabled
l2ptguard                    Disabled
link-flap                    Enabled          
mac-limit                    Disabled
link-monitor-failure         Disabled
loopback                     Disabled
oam-remote-failure           Disabled
pagp-flap                    Enabled
port-mode-failure            Disabled
psecure-violation            Enabled
security-violation           Enabled
sfp-config-mismatch          Disabled
storm-control                Disabled
udld                         Enabled
vmps                         Enabled

Timer interval: 60 seconds                     

Interfaces that will be enabled at the next timeout:

Interface       Errdisable reason       Time left(sec)
---------       -----------------       --------------
Gi1/6                 link-flap           47

Root cause for the problem with interface Gi1/6 lies in link-flap. This is somewhat cryptic but will give you an idea where to find the solution.

Possible root causes are;
dodgy cabling
faulty optics (sfp/gbic)
wrong speed/duplex settings
hardware loop

In this particular case I suspected the speed/duplex settings.

somerouter(config-if)#speed nonegotiate
somerouter(config-if)#no shut
Jan 10 11:20:31.589 CET: %LINK-5-CHANGED: Interface GigabitEthernet1/6, changed state to administratively down
Jan 10 11:20:33.703 CET: %LINK-3-UPDOWN: Interface GigabitEthernet1/6, changed state to up
Jan 10 11:20:33.712 CET: %LINEPROTO-5-UPDOWN: Line protocol on Interface GigabitEthernet1/6, changed state to up

Link was forced down and up to get rid of the error-disable condition.
Terminal monitor was turned on to instantly see what is going on.
Logging tells me that the port is admin down and that changed to up.
Et voilà, line protocol is also changed to up.

Root cause found and resolved.

Posted in CCNP | Tagged , ,

SSH error on cisco router

Enabling SSH on a Cisco router is not enough to guarantee SSH is working.
Enabling ssh version 2.0 on a routers is as straight forward as;.

>ip ssh version 1-2

I surely recommend the use of version 2, because ssh v1 has a bug which allows root access.

Anyhow… ssh needs some sort of encryption to work. In other words create a crypto key.

host(config)#crypto key generate rsa modulus 2048
The name for the keys will be: host.somenetwork.com
% The key modulus size is 2048 bits
% Generating 2048 bit RSA keys, keys will be non-exportable...
[OK] (elapsed time was 2 seconds)
Nov  5 14:17:17.240 UTC: %SSH-5-ENABLED: SSH 2.0 has been enabled

The good thing is the crypto key is not saved in config. The bad thing is, whenever you commission a router and forget to create the crypto key it can cost you some time in troubleshooting. Worst case you have to travel on-site to create the crypro key.

By the way, make sure you enable ssh on the VTY ports and if applicable deny telnet input.

line vty 0 15 
transport input ssh
no transport input telnet

And create an access-list and apply it to the VTY ports.

FYI: SSH enables encrypted CLI access towards a host. And the crypto key validates the host communication i.e. detection of a man-in-the-middle attack.

Posted in CCNP

DHCP Helper address

This week a project manager contacted me with an rather peculiar question. I work with this project manager now for almost two years. And I know he is technically very skilled.  So if he doesn’t know the answer then the answer is not always straight forward.

I currently work for a large telco in the Netherlands. One of the services offered is DHCP-relay. Technically this is a very basic and straight forward service. The customer wanted to know which address is used as source address in the packets destined to the DHCP server. This was needed to set some static routes.

Okay lets start labbing, within GNS3 I made the following lab setup;


IP addresses used are;

  • DHCP-Server
    • e0/0
  • Router
    • e0/0
    • e1/0.1 (switch management vlan)
    • e1/0.100 (customer vlan)
  • Switch
    • vlan1
  • PC1
    • e0 will get ip from DHCP server.

Router config  interface facing the dhcp client.

interface Ethernet1/0.100
 encapsulation dot1Q 100
 ip address
 ip helper-address  

With the set up ready I am able to get an ip address. During the DHCP negotiation wireshark is gathering information on the link between the router and the dhcp-server.

DHCP request and reply

Opcode: 1 (REQUEST)
Client IP Address:
Your IP Address:
Server IP Address:
Gateway IP Address:
Client MAC Address: 00:50:79:66:68:00
Option 53: Message Type = Request
Option 54: DHCP Server =
Option 50: Requested IP Address =
Option 61: Client Identifier = Hardware Type=Ethernet MAC Address = 00:50:79:66:68:00
Option 12: Host Name = PC11

Some messages have been omitted for readability. In the process prior to assigning an IP address the DHCP server chooses an IP and checks uniqueness in the network using ping.

Opcode: 2 (REPLY)
Client IP Address:
Your IP Address:
Server IP Address:
Gateway IP Address: 
Client MAC Address: 00:50:79:66:68:00
Option 53: Message Type = Ack
Option 54: DHCP Server =
Option 51: Lease Time = 86400
Option 58: Renewal Time = 43200
Option 59: Rebinding Time = 75600
Option 1: Subnet Mask =
Option 3: Router =


To be honest I set a static route on the DHCP-server

ip route

What does Wireshark tell me ?  This whole exercise is undertaken to ‘see’ what information passes from the router to the dhcp-server and vice-versa.

I expect a dhcp discover followed by a offer. Between discover and offer the dhcp-server should ping the ip address which it is going to serve to the dhcp-client. No ping reply means; ip address if not currently in use. No duplicate ip addresses are injected.


A little close-up reveals the answer;


The DHCP unicast packet is sourced by which is the router interface facing the dhcp-client. This interface contains the ip helper-address statement.

NOTE: the network diagram is overly simplified. First because I do not wish to reveal any business proprietary information. Second it withdraws focus from the question at hand.

NOTE2: Thank you David Theuns for being your ‘vraagbaak’.

Posted in CCNP, GNS3 | Tagged , , , | 1 Comment

GNS3 caveat – port status

Creating a network lab in GNS3 is reliefs on budget. But there are some drawbacks. One of which is port status. I have noticed that using GNS3 in combination with iouyap results in some unexpected behavior.

Once a topology is made in GNS3. And one interface is set administrative operational, with no shut. The port immediately becomes operational i.e. up/up. Within GNS3 there is no such thing as admin up/operational down.

At this point I know that two protocols can help detect L2 communication with adjacent node. Those are UDLD and BDF.

Beware not to waste time in resolving/troubleshooting GNS3 problem. Instead of resolving network problems.

Posted in Commands, GNS3 | Tagged , , ,

Extended IP Access-list

As we all have learned in our pursuit to CCNA there are many different types of access-list. In reality most commonly used are ip access lists.

Major difference between ip standaard access-list an ip extended access-list is the ability to insert and remove entries within the ACL. When an entry has to be removed with in an standard ACL the entire ASL need to be removed and added again. The latter off course without the ACE which needs removal.

A extended ACL has slightly more steps but has some major benefits. Let say you have the following ACL;

access-list 130 permit ip host host
access-list 130 permit ip host any
access-list 130 deny ip any any

And the entry “permit ip host any” must be replace with “permit ip host host”

The following steps need to be taken are;

  • Obtain information about the ‘rule number’
  • Insert new ACE
  • Remove wrong ACE
  • Resequence rule numbers

Obtain rule number information

R4#sh access-lists 130
Extended IP access list 130
 10 permit ip host host
 20 permit ip host any
 30 deny ip any any

Insert new ACE

Rule 11 until 19 are free to use. Same goes for 21 until 29. Let use number 15.

*Feb 19 13:13:35.871: %SYS-5-CONFIG_I: Configured from console by console
R4#conf t
Enter configuration commands, one per line. End with CNTL/Z.
R4(config)#ip access-list extended 130 
R4(config-ext-nacl)#15 permit ip host host
*Feb 19 13:14:24.206: %SYS-5-CONFIG_I: Configured from console by console
R4#sh access-lists 130 
Extended IP access list 130
 10 permit ip host host
 15 permit ip host host
 20 permit ip host any
 30 deny ip any any

Remove wrong ACE

ACE that needs removal has rule number 20. Removal is as easy as “no 20” within the access-list.

R4#conf t
Enter configuration commands, one per line. End with CNTL/Z.
R4(config)#ip access-list extended 130 
R4(config-ext-nacl)#no 20
*Feb 19 13:18:32.452: %SYS-5-CONFIG_I: Configured from console by console
R4#sh ip access-lists 130 
Extended IP access list 130
 10 permit ip host host
 15 permit ip host host
 30 deny ip any any

The rule numbers are somewhat borked. This needs a little repair.

Resequence rule numbers

Repair command is;

ip access-list resequence

*Feb 19 13:22:42.838: %SYS-5-CONFIG_I: Configured from console by console
R4#conf t
Enter configuration commands, one per line. End with CNTL/Z.
R4(config)#ip access-list resequence 130 10 10
*Feb 19 13:23:12.473: %SYS-5-CONFIG_I: Configured from console by console

R4#sh access-list 130 
Extended IP access list 130
 10 permit ip host host
 20 permit ip host host
 30 deny ip any any

Naturally the most important benefit is that the access-list is applied all the time. In regard of the standard ip access-list; a slight modification results in removing the ACL and re-instating the ACL again. You will never know which traffic passes the interface while working on the access-list.

The same applies for named access-list.


Cisco NEXUS devices have a slight different syntax to resequence
NEXUS-R3(config)# resequence ip access-list some_vlan 10 10
Posted in CCNP, Commands | Tagged , , , ,

EIGRP neighbors

Cisco EIGRP protocol allows different methods to form neighborship relations. Default method is form neighborship automatically via multicast (multicast address of is used)

After an adjacency is formed ‘hellos’ are  sent using IP protocol 88 (eigrp). This means that;

  1. an interface need to be IP enabled
  2. both interfaces to a link need to have an ip address within the same subnet.

Alternative method of creating an neighborship relation is setting the neighborship ‘hard-coded’. This is done withing the eigrp routing context. Setting a neighborship ‘over’ an interface implicitly disables the ability to form adjacencies using multicast. Inbound EIGRP multicast packets are discarded and outbound are not send out that particular interface.

This alternative method can be used when you want to prevent a rogue router in your network inject all sorts of garbage subnets.

Off course a better way to set a md5 key on eigrp update exchanges.

Posted in Commands