Hi everybody.
Here I am again with another silly scenario.
If one of the link in etherchannel goes down, then Pagp or LCAP simply shifts the load from failed link to other active links in etherchannel.
The question is should we expect the same behavior if etherchannel is configured statically using channel-protoco 1 mode on ?
Below is my lab set up:
host 2------SW1 f1/1---------------f1/1 SW2-----host4
f1/2---------------f1/2
f1/1 and f1/2 on both switches are bundled in etherchannel which is acting as a trunk.
hos2 ip 199.199.199.2
host4 ip 199.199.199.4
Sw1 has default load balance method i.e src-dst -ip for etherchannel
etherchannel is configured statically with :
channel -protocol mode on on both switches.
Sw1 and Sw2 assign port index as:
port index 0 f1/1
port index 1 f1/2
Let say host2 sends a ping packet to host4 ( 199.199.199.4).
Focusing only on the path this packet will take, following will occur:
Sw1 receives the packet and Perform XOR operation on the right most bits of src ip( 199.199.199.2) and destination ip ( 199.199.199.4)
.
the last octet of 199.199.199.2 is:
.2 --> 0 0 0 0 0 010
The last octet of 199.199.199.4 is:
.4---> 0 0 0 0 0100.
Performing XOR operation on the right most bit will yield
0 XOR 0= 0
So sw1 will use link f1/1 because port index assigned to it in etherchannel is 0 to forward ping packets.
Now what would happen link f1/1 for some reason goes down, Will sw1 drop ping packet sourced from 199.199.199.2 and destined to 199.199.199.4 ?
Will sw1 simply use the other link i.e f1/2 to foward them?
I simulated the above scenario by shutting down f1/1 on sw1. First let me show you the output before I disabled the portf1/1 on sw1.
( I am using router for host below)
host2#ping 199.199.199.4
Type escape sequence to abort.
Sending 5, 100-byte ICMP Echos to 199.199.199.4, timeout is 2 seconds:
!!!!!
SW1 output:
SW1#show etherchannel port
Channel-group listing:
-----------------------
Group: 1
----------
Ports in the group:
-------------------
Port: Fa1/1
------------
Port state = Up Mstr In-Bndl
Channel group = 1 Mode = On/FEC Gcchange = 0
Port-channel = Po1 GC = 0x00010001 Pseudo port-channel = Po1
Port index = 0
Age of the port in the current state: 00d:00h:01m:51s
Port: Fa1/2
------------
Port state = Up Mstr In-Bndl
Channel group = 1 Mode = On/FEC Gcchange = 0
Port-channel = Po1 GC = 0x00010001 Pseudo port-channel = Po1
Port index = 1
Age of the port in the current state: 00d:00h:05m:50s
====================================================================
Everything is working normal thus far.
Now lets disables the port f1/1 on sw1.
SW1(config)#int f1/1
SW1(config-if)#shutdown
*Mar 1 00:22:23.991: %LINK-5-CHANGED: Interface FastEthernet1/1, changed state to administratively down
Sw1#show etherchannel summary
Flags: D - down P - in port-channel
I - stand-alone s - suspended
R - Layer3 S - Layer2
U - in use
Group Port-channel Ports
-----+------------+-----------------------------------------------------------
1 Po1(SU) Fa1/1(D) Fa1/2(P)
The port f1/1 is down
Next , we have h2 send ping packet and let see what happens.
host2#ping 199.199.199.4
Type escape sequence to abort.
Sending 5, 100-byte ICMP Echos to 199.199.199.4, timeout is 2 seconds:
.!.!.
Success rate is 40 percent (2/5), round-trip min/avg/max = 1572/1744/1916 ms
===============================================================
1)Based on the above result, it appears to me sw1 does shift the load from failed link to other active links in etherchannel. Is it true or Is it some bug in IOS? ( keep in mind the etherchannel in our case is staticallly configured using channel-group 1 mode on)
2) If you look at the output below
.!.!.
First packet no response
second packet response
third packet no response
fouth packet response
fifth packet no response
I do not understand why we are getting response for some pings packets and no response for others?
thanks and have a great day.
