virtual cat 9k with CML/DNAC

[Mike McFarlin]

Is there any way of randomizing the hardware serial number of a virtual cat 9k in CML?

I'm trying to lab DNAC with a pair of 9ks. DNAC won't complete the import of the second 9k because of serial number conflict.

Thanks,

Mike McFarlin

[Mike McFarlin]

Found the answer on the serial number. I'm wondering how other people got past this in DNAC. 2.3.3.7 is choking on it.

https://developer.cisco.com/docs/modeling-labs/#!cat-9000v/cat-9000v

[zooapk091]

Changing or randomizing the hardware serial number of a virtual Cisco Catalyst 9k (Cat 9k) device in Cisco Modeling Labs (CML) can be a bit tricky. The hardware serial number is typically a unique identifier assigned to physical devices and is not something you would normally modify for virtual devices. However, for lab and testing purposes, you might want to find a workaround to avoid conflicts like the one you mentioned with DNAC.

Here are a few suggestions:

1. Clone the Virtual Machine (VM): If you're running the Cat 9k in a virtualized environment, you can create a clone of the VM, which should give you a new unique identifier. Ensure that you configure the cloned VM appropriately.

2. Contact Cisco Support: If you have a legitimate need to change the serial number, you might want to reach out to Cisco support for guidance. They might be able to provide a solution or workaround specific to your situation.

3. Use Different Lab Environments: Instead of running both Cat 9k instances in the same CML lab environment, you could create separate lab environments for each device. This way, they won't conflict with each other's serial numbers.

4. Check DNAC Settings: Double-check your DNAC settings and ensure that there are no configuration errors causing the conflict. Sometimes, the issue might not be related to the serial number but rather a configuration issue.

5. Virtual MAC Address Changes: In some virtualization environments, you can change the MAC address of the virtual network adapter, which could indirectly affect the serial number conflict. However, this is a workaround and may not be a recommended practice.

Remember that changing or randomizing hardware serial numbers can have unintended consequences, so proceed with caution and ensure that it aligns with your lab and testing goals. Always follow best practices and adhere to licensing agreements when working with virtualized network devices.

[Mike McFarlin]

Cloning the node didn’t change the serial number. I believe the serial number is built into the node image.

[Torbjørn]

@Mike McFarlin, did you figure out a sensible way to do this?

[Mike McFarlin]

Unfortunately no. I hope this becomes a usable option. But I’m not waiting for it. I’m going with 3850 for my DNAC lab.

[Torbjørn]

That's a good way to go. My SDA lab setup currently consists of ISR4331s and a mix of 3650, 3850 and 9300.

It's just so much more practical with virtual appliances... Rumor has it that there is a version of the cat9kv that isn't suffering from this issue. Hopefully it will be included in the next version of the refplat.

[timshotton]

I don't suppose you know if that rumour is any truer since you first posted this do you? I've just run into this same issue after getting all excited to build a fabric with these nodes haha

[Torbjørn]

Terry Vinson demoes it in the following video: Build Your 100% Virtual CCIE Enterprise v1.1 Home Lab With Cisco CAT9KV And DNAC. It is apparently a _very_ early build of IOS-XE 17.13. So hopefully it will be made generally available soon.

Did you get any useful response from your inquiries @Arne Bier? ref. your learning community post.

[timshotton]

Oh nice one, thanks for that link! I've seen he's dropped a link to a qcow image that might work.. Downloading it now and will report back!

[Arne Bier]

Hey folks,

I only got as far as importing one CAT9KV into my lab DNAC. I didn't know about all this serial number shenanigans until I saw it mentioned. And I also gave up trying to do anything with this beta CAT9KV because it doesn't do the basics - like MAC learning. Don't try and build any kind of lab with it that involves a learning path or certification path. This thing will lead you up the garden path!

The version I have is 17.10.something - it was compiled in Sept 2022 - it no longer supports the UADP option - only Silicon One. Not that that matters - but in the Cisco Live Preso material they went into great depths about these "options". 

@Torbjørn I have reached out multiple times to the Cisco Learning Network, I even emailed CML, as well as one of the CML presenters from Cisco Live 2023 in Melbourne, and no response from his Webex Teams Room. It's ridiculous. No reply at all. At this point I feel like something is deeply wrong with whoever is in charge of that "product" - either they no longer work at Cisco and Cisco is in trouble, or they realised they made a mistake and don't want to update or release another version.

I hope by me saying this out in public, I jinx it and am proven wrong - I hope I am wrong. Because this C9KV is a great idea and I wish it would work. 

The thing that grinds my gears annoys, is Cisco's lack of professional courtesy of not even responding to any of my many attempts to just get a basic answer.

[bigevilbeard]

@Arne Bier  The best folks to ask would be Joe Clarke or Ralph Schmieder. Both are super helpful and we’re always my go to for CML. 

Hope this helps.

[Torbjørn]

I reached out to Joe Clarke and was told that a new 17.12 image will be released with CML 2.7 in February.

[timshotton]

Oh sweet! I've managed to get the image that Terry listed in the comments of his Youtube video.. I've got them added to DNAC now, but a bit hampered by this 256Kbps throughput on these nodes.. I have an edge connected to a border and the ping response times are sketchy.. Seeing if it will provision and be added to the fabric with an edge role now.. 

[Arne Bier]

Yep - I think I jinx'd it. At the same time I got a reply from a bunch of Cisco people and they told me the same thing.

And they also told me MAC address learning is unsupported. I still can't believe it. A switch that does not support MAC address learning is not much of a switch in my opinion.