Traceroute - Cisco, Linux, and Windows versions

In my today post I’d like analyze different version of traceroute on three platforms. You need to know exactly what protocols/types are used when you would like to permit them on your firewall.

1) Cisco version

traceroute-cisco-1.jpg

I’m going now traceroute from R17 to R18:

traceroute-cisco-1a.jpg

This is what I captured on R17 interface:

traceroute-cisco-2.jpg

Let me explain what we see:

R17 sends udp packet with ttl=1 to discover device in one hop distance. R16 decrements ttl by 1 and sees that ttl=0 and sends icmp packet ‘time exceeded):

-> udp - dst port: 33434, ttl=1
<- icmp - time exceeded (due to ttl=0) - type 11, code 0 - ttl=255

-> udp - dst port: 33435, ttl=1
<- icmp - time exceeded (due to ttl=0) - type 11, code 0 - ttl=255

-> udp - dst port: 33436, ttl=1
<- icmp - time exceeded (due to ttl=0) - type 11, code 0 - ttl=255

After three repeats R17 increases ttl by 1 and sends next three packets. R15 receives them, decreases ttl by 1 and due to ttl=0 R15 sends back icmp 'time exceeeded’:

-> udp - dst port: 33437, ttl=2
<- icmp - time exceeded - type 11, code 0 - ttl=254

-> udp - dst port: 33438, ttl=2
<- icmp - time exceeded - type 11, code 0 - ttl=254

-> udp - dst port: 33439, ttl=2
<- icmp - time exceeded - type 11, code 0 - ttl=254

After next three packets R17 increases ttl by 1 to ttl=3. These three packets reach R18 and due to invalid udp port (range 33434-33464) R18 sends icmp message: port unreachable (type:3, code:3):

-> udp - dst port: 33440, ttl=3
<- icmp - port unreachable - type 3, code 3 - ttl=253

-> udp - dst port: 33441, ttl=3
<- icmp - port unreachable - type 3, code 3 - ttl=253

-> udp - dst port: 33442, ttl=3
<- icmp - port unreachable - type 3, code 3 - ttl=253

On firewall we need to open following (returning) packets: icmp time-exceeded (type 11, code 0) and port unreachable (type 3, code 3).

2) Linux version

I added Linux machine to my network:

traceroute-linux-1.jpg

and I traceroute from Linux to r18:

traceroute-linux-2.jpg

Below you can see what I captured:

traceroute-linux-3.jpg

Let’s analyze what we see:

Linux sends udp packet with ttl=1 to discover device in one hop distance. R17 decrements ttl by 1 and sees that ttl=0 and sends icmp packet ‘time exceeded):

-> udp - dst port: 33434, ttl=1
<- icmp - time exceeded (due to ttl=0) - type 11, code 0 - ttl=255

-> udp - dst port: 33435, ttl=1
<- icmp - time exceeded (due to ttl=0) - type 11, code 0 - ttl=255

-> udp - dst port: 33436, ttl=1
<- icmp - time exceeded (due to ttl=0) - type 11, code 0 - ttl=255

After three repeats Linux increases ttl by 1 and sends next three packets. R16 receives them, decreases ttl by 1 and due to ttl=0 R16 sends back icmp 'time exceeeded’:

-> udp - dst port: 33437, ttl=2
<- icmp - time exceeded - type 11, code 0 - ttl=254

-> udp - dst port: 33438, ttl=2
<- icmp - time exceeded - type 11, code 0 - ttl=254

-> udp - dst port: 33439, ttl=2
<- icmp - time exceeded - type 11, code 0 - ttl=254

After three repeats Linux increases ttl by 1 (to 3) and sends next three packets. R15 receives them, decreases ttl by 1 and due to ttl=0 R15 sends back icmp ‘time exceeeded’:

-> udp - dst port: 33440, ttl=3
<- icmp - time exceeded - type 11, code 0 - ttl=253

-> udp - dst port: 33441, ttl=3
<- icmp - time exceeded - type 11, code 0 - ttl=253

-> udp - dst port: 33442, ttl=3
<- icmp - time exceeded - type 11, code 0 - ttl=253

In the last step Linux sends 7 udp packets with ttl=4 (3 packets), ttl=5 (3 packets) and ttl=6 (1 packet). Every packet generates responds: icmp port-unreachable.

-> udp - dst port: 33443, ttl=4
<- icmp - port unreachable - type 3, code 3 - ttl=252

-> udp - dst port: 33444, ttl=4
<- icmp - port unreachable - type 3, code 3 - ttl=252

-> udp - dst port: 33445, ttl=4
<- icmp - port unreachable - type 3, code 3 - ttl=252
-> udp - dst port: 33446, ttl=5
<- icmp - port unreachable - type 3, code 3 - ttl=252

-> udp - dst port: 33447, ttl=5
<- icmp - port unreachable - type 3, code 3 - ttl=252

-> udp - dst port: 33448, ttl=5
<- icmp - port unreachable - type 3, code 3 - ttl=252
-> udp - dst port: 33449, ttl=6
<- icmp - port unreachable - type 3, code 3 - ttl=252

As you see more or less the variation is similar to the Cisco one. I can’t find any information why Linux (2.6.32) needs to send last 7 packets instead of 3.

3) Windows version

traceroute-windows-1.jpg

For this version I use the same scenario as for linux.

traceroute-windows-2.jpg

The Windows version uses only icmp packets (no udp). Below you can see what I captured:

traceroute-windows-3.jpg

-> icmp - echo request, ttl=1
<- icmp - time exceeded (due to ttl=0) - type 11, code 0 - ttl=255

-> icmp - echo request, ttl=1
<- icmp - time exceeded (due to ttl=0) - type 11, code 0 - ttl=255

-> icmp - echo request, ttl=1
<- icmp - time exceeded (due to ttl=0) - type 11, code 0 - ttl=255

traceroute-windows-4.jpg

-> icmp - echo request, ttl=2
<- icmp - time exceeded (due to ttl=0) - type 11, code 0 - ttl=254

-> icmp - echo request, ttl=2
<- icmp - time exceeded (due to ttl=0) - type 11, code 0 - ttl=254

-> icmp - echo request, ttl=2
<- icmp - time exceeded (due to ttl=0) - type 11, code 0 - ttl=254

traceroute-windows-5.jpg

-> icmp - echo request, ttl=3
<- icmp - time exceeded (due to ttl=0) - type 11, code 0 - ttl=253

-> icmp - echo request, ttl=3
<- icmp - time exceeded (due to ttl=0) - type 11, code 0 - ttl=253

-> icmp - echo request, ttl=3
<- icmp - time exceeded (due to ttl=0) - type 11, code 0 - ttl=253

traceroute-windows-6.jpg

-> icmp - echo request, ttl=4
<- icmp - echo reply - type 11, code 0 - ttl=252

-> icmp - echo request, ttl=4
<- icmp - echo reply - type 11, code 0 - ttl=252

-> icmp - echo request, ttl=4
<- icmp - echo reply - type 11, code 0 - ttl=252

For Windows version, on a firewall we need to open: icmp time exceeded and echo-reply (in case we don’t inspect icmp traffic).

 
5
Kudos
 
5
Kudos

Now read this

DMVPN - phase one - EIGRP

Today I would like to implement DMVPN with EIGRP. This protocol is very popular because of its scalability. Please read this post before you start because I’m not going to implement it from scratch:... Continue →