Fast convergence vpc.

[Mlex1]

Hi 

fast-convergence am i right we use this command on vpc when link between peers flapped, i searched a few links especially on cisco website.

and also found this.

Fast Convergence is used when the peer-link flaps. When the peer-link flaps, the secondary may shutdown ports.

By default, it shuts down VLAN SVI's faster than it shuts down physical ports. Once again, this results in traffic being blackholed.

The new feature shuts down the ports and the SVI's at the same time, preventing blackholing.

Apparently these features can cut 500ms of loss down to 50-250ms.

https://www.reddit.com/r/Cisco/comments/76un42/vpc_fast_convergence/ 

[M02@rt37 · Accepted solution]

Hello @Mlex1 

The vPC Fast Convergence feature is a critical enhancement in environments where minimizing traffic loss during failure scenarios is essential.

It is particularly useful when the vPC peer-link flaps. In such cases, the default behavior of a secondary vPC switch involves shutting down VLAN SVI interfaces faster than physical vPC member ports. This behavior can cause traffic to be blackholed temporarily, as physical ports continue forwarding traffic while the SVIs are no longer active.

By enabling the vPC Fast Convergence feature, this issue is addressed by ensuring that both SVIs and physical vPC member ports are shut down simultaneously when a peer-link failure is detected. This synchronization prevents traffic from being blackholed, as the entire path is effectively removed from forwarding at the same time.

The implementation of vPC Fast Convergence is designed to significantly reduce traffic loss during such events. It can cut downtime from approximately 500 milliseconds to a range of 50 to 250 milliseconds, which is critical for applications requiring high availability and low latency. So this feature is particularly valuable in DC environments where vPC is heavily relied upon for redundancy and load balancing...

Note that "fast-convergence" command does not exist on Nexu 7000, however if you enable "vpc orphan-port suspend" you will get the same behavior.

[Mlex1]

what about  LACP convergence same result when we active?

[M02@rt37]

Hello @Mlex1 

When LACP is active, it enables dynamic link monitoring and automatic adjustments within a port-channel, ensuring rapid convergence during link failures or recoveries. By exchanging LACPDU packets, it quickly detects and removes failed links from the aggregation group, redistributing traffic across the remaining operational links. If a link recovers, LACP dynamically reintegrates it into the group. This process minimizes traffic disruption and eliminates blackholing, achieving fast convergence typically within milliseconds to seconds, depending on network conditions, making it significantly more reliable than static link aggregation.

[Mlex1]

MO2@rt37 as i mansion above, thank you for quick support wish you all the best 

The new feature shuts down the ports and the SVI's at the same time, preventing blackholing.

[Mlex1]

after enabling fast convergence on vpc peers, periodically notice this log, on eve-ng

SW-01 %$ VDC -1 %$ % VPC -2-PEER-KEEPALIVE RECV FAIL: In domain 1, VPC peer keep-alive receive has failed

[MHM Cisco World]

I think I know what happened here but you can confirm that

1- you test fast converge via power off the primary NSK ? if Yes then your test is wrong you need to shut the peer-link only 
OR 
2-fast converge is make secondary immediate shut the port and down the SVI, I think here you use VLAN SVI for peer-keepalive 
and then allow this VLAN in peer-link so now this VLAN is become vpc vlan 
when you shut the peer-link in primary the secondary immediate down vpc vlan include the vlan you use for keep-alive, this make secondary and primary show log of peer-link failed 

MHM

[Mlex1]

Hello @MHM Cisco World 

1- you test fast converge via power off the primary NSK ? if Yes then your test is wrong you need to shut the peer-link only 
OR 

not

but i will do what you typed upper