I lied.  I am working on IPv6 now.  I’ll do IP Services later.  I am pretty happy with IPv6.  Unfortunately the NM-16-ESWs in my 3640s on Dynamips have a fart when you try and configure an L3 Etherchannel between them.  Now I dont know if this is a Dynamips specific drama, if its the IOS I’m using, or if its the NM-16ESW itself.

*Mar  1 01:22:08.799: %LINEPROTO-5-UPDOWN: Line protocol on Interface FastEthernet0/15, changed state to up

SW-1(config-if-range)#channel-group 1 mode on

SW-1(config)#interface range f0/14 - 15
SW-1(config-if-range)#no switchport
*Mar  1 01:21:46.411: %LINEPROTO-5-UPDOWN: Line protocol on Interface FastEthernet0/14, changed state to up
*Mar  1 01:21:46.495: %LINEPROTO-5-UPDOWN: Line protocol on Interface FastEthernet0/15, changed state to up

would not accept channel-group command with no switchport

SW-1(config-if-range)#switchport
SW-1(config-if-range)#channel-group
*Mar  1 01:22:08.763: %LINEPROTO-5-UPDOWN: Line protocol on Interface FastEthernet0/14, changed state to up

Creating a port-channel interface Port-channel1

OK, L2 channel seems ok

SW-1(config-if-range)#
*Mar  1 01:22:15.135: %EC-5-BUNDLE: Interface Fa0/14 joined port-channel Po1
*Mar  1 01:22:15.171: %EC-5-BUNDLE: Interface Fa0/15 joined port-channel Po1
SW-1(config-if-range)#no switchport

Lets try making the channel group then making it L3 after

SW-1(config-if-range)#
*Mar  1 01:22:18.091: %LINEPROTO-5-UPDOWN: Line protocol on Interface Port-channel1, changed state to up
*Mar  1 01:22:18.571: %EC-5-UNBUNDLE: Interface Fa0/14 left the port-channel Po1
*Mar  1 01:22:18.575: %ESWILP_FLTMG-7-INTERNAL_ERR: Internal error: *** failure to create entry in vtable/vlan 1006/unit 0 -Traceback= 0×603C5124 0×62445AD0 0×6242F2D4 0×62435C50 0×62446340 0×6047F6C4 0×62439F9C 0×6240F680 0×624176B0 0×604057D8 0×604218B0 0×604C229C 0×604C2280

ARRRGGGGHHHHHHHHH

SW-1(config-if-range)#
*Mar  1 01:22:18.595: %EC-5-UNBUNDLE: Interface Fa0/15 left the port-channel Po1
*Mar  1 01:22:19.595: %LINEPROTO-5-UPDOWN: Line protocol on Interface Port-channel1, changed state to down

Up yours Cisco!!!

Played around with it for a bit… Traceback city no matter what I did.  Im not that fussed though, Im pretty happy with RIPng anyway.  I already know how to make L3 Etherchannels so its all good.  I might come back later and just do some inter-router RIPng stuff rather than trying to do it over L3 Etherchannels.  It’s no different really.

Narbik’s OSPFv3 labs are just router only… Let’s try those.

I finished Narbik’s book today. Pretty happy with that I might add. Certainly picked up a few gotchyas on the way through there. I just hope I remember them all for the upcoming months. Tomorrow I will go over the things I had trouble with on the InternetworkExpert CoD. Then I will repeat those sections of soup-to-nuts again and see if I do better.

One thing that has an incredibly high suck factor is the fact IPv6 does NOT WORK over frame-relay on my lab setup.  I tried various methods of configuring (point-to-point, multipoint, frame maps) and nothing.  I’m not that fussed though…  I went through the answers thinking “I would type this line here” and there they were were I wanted them.  A few things were different but thats all good.  I havent done enough testing to see if it is a problem with the IOS I am using or if it is something else.  My guess is IOS rather than being a dynamips/GNS3 problem because the frame mappings work fine with IPv4, as well as running IP across the links.  So Im saying IPv6 IOS bug…

Im going to be generous once again and include the startup-configs I rewrote to work on my dynamips setup.  I improved the configs as I went through the lab so some of them might not be a straight copy and paste, also some labs I did not edit as they had the same initial config as a previous lab.  You will find these here and there.  So if these things save you about 15-30 mins on the start of 90% of the labs I am sure you wont be complaining to me too much

initial-configs

Now to open a bottle of some fine Barossa Valley Shiraz.  mmmm

A few people have asked for my .net file for Narbik’s topology. Be easier to put it here than email everyone who wants it. This topology matches the Soup-to-nuts topology with a few differences:

Switches are really 3640s with NM-16ESWs in them - remember hard code duplex settings, and lacking all switch features
All of Narbik’s FE interfaces are now Ethernet, so when you copy the startup configs, just edit them to reflect E0/[01] rather than F0/[01]
Serial interfaces are in S1/0 rather than S0/0. Once again modify startup configs to suit, also remember to hard code lmi-type to ansi for it to work.

Remember to set the idle-pc values for your own setup so your CPU doesnt go through the roof.

Feel free to use the file, and let me know what you think!

narbik.net

Dynamips is a great tool. Not everyone can afford to spend $15k or whatever on a rack with real routers but with Dynamips you can get most of the functionality out of a $750 PC.

I went down to MSY recently and got a PC to do the job. You need lots of RAM (4GB for my box), a beefy CPU (2.66GHz Core2Duo) and you’re on your way. Windows is a pain for Dynamips for a variety of reasons. I am using Ubuntu 7.10 as my OS. I am a Mac user normally but I couldnt be bothered trying to get OSX running on a yum-cha beige box when I can get Linux running in about 10 minutes…

GNS3 is the GUI version of Dynamips. I have made a Narbik topology in GNS3 and it does the job very well. Pitfalls are:

No Switch emulation. I run a 3640 with an NM-16ESW as a switch. It does probably half the job a real 35[56]0 does. No MST, no auto vlan creation from the interface (must use vlan database first), no vlan creation from conf t (once again use vlan database) and also cant auto detect duplex settings. I got around this by hard coding the duplex on the connected router ports.

Frame-relay wont auto-detect lmi-type. Simple… frame-relay lmi-type ansi.

3640s arent used in the real lab. No big deal. The IOS is the same (almost)… the only difference is interface numbering. What you need to watch out for though is the NM-16ESWs are only Ethernet and you cant get FE or GE blades for the 3640s. This affects spanning-tree costs and the like but its no big deal. If you are really pedantic you could manually asssign spanning-tree port costs and the like to interfaces… But I didnt bother unless my scenario asked me to.

This is my first post of my under construction CCIE blog. Speaking to Arden convinced me to start so here I go.

I am studying for my CCIE (Routing and Switching) at the moment. The materials I am using are:

InternetworkExpert Workbooks and Class on Demand. The CoDs are FANTASTIC and are a requirement if you wish to pursue your CCIE.
Narbik’s Soup-to-Nuts book
Micronics bootcamp - May 2008 in Sydney
Dynamips PC - 4 GB RAM, Ubuntu 7.10, GNS3

Feel free to post comments!