Does turning off MCP cause downtime?

[Network-admin123456]

Hi all,

I have an issue where I have exceeded the 2000 VLANs + port MCP limit on multiple Leaf switches. I am considering turning off MCP towards non loop inducing devices, such as bare-metal servers and L3 devices to help mitigate this limit. My question is will turning off MCP on a number of ports cause downtime towards those hosts? Or is there perhaps a better way to mitigate this limit?

Also, would MCP STRICT mitigate or better scale this issue? Currently on 5.2(3) so would need to upgrade to 5.2(4) to get MCP STRICT feature. I have not found what is the VLAN + port limit on MCP STRICT.

I have not found anything in ACI best practices that would give me a decisive answer to this question, hence I am asking the community.

Regards

[AshSe]

Before coming to the conclusion, let's understand the role of MCP:

• MisCabling Protocol (MCP) can detect an external Layer 2 forwarding loop by non-direct link (via external L2 switch)
• MCP frames is sent on all VLANs, on all ports.
• if any switch detects MCP packet arriving on a port that originated from the same fabric the port is put to error disabled state.

Look at the diagram:

 Now a question to you:

Q. Have you configured MCP Instance globally?

If the answer is yes, then you should not remove MCP.

if the answer is no, you can remove MCP from the selected ports.

HTH

[Network-admin123456]

Hi,

MCP is globally enabled. What would happen if I were to disable MCP from a single port, given that MCP is globally enabled?

Regards

[AshSe]

Turning off MCP (MisCabling Protocol) on a single interface in Cisco ACI is generally less risky than disabling it globally, but it can still potentially cause network issues, including downtime, depending on your network topology and configuration. Here are some considerations:

1. Risk of Loops: MCP is designed to detect and mitigate Layer 2 loops. Disabling it on a single interface means that if a loop occurs on that interface, it won't be detected and mitigated by MCP, which could lead to broadcast storms and network instability.

2. Network Design: If your network is well-designed and you are confident that there are no potential loops involving that interface, the risk is lower. However, it's essential to thoroughly understand your network topology before making such changes.

3. Impact Scope: The impact of disabling MCP on a single interface is localized to the segment of the network connected to that interface. However, issues in one part of the network can sometimes propagate and affect other areas.

4. Maintenance Window: As with any network change, it's best to perform this action during a maintenance window to minimize potential impact on users.

5. Monitoring: After making the change, closely monitor the network for any signs of issues. Be prepared to revert the change if problems arise.

6. Documentation and Communication: Document the change and communicate with relevant stakeholders to ensure everyone is aware of the potential risks and the reasons for the change.

In summary, while turning off MCP on a single interface is less risky than disabling it globally, it still carries some risk. Careful planning, understanding your network topology, and monitoring are essential to minimize potential issues. If in doubt, consulting with a network engineer or Cisco support is advisable.