Configuring Optimized BFD Path
To address BFD flaps, NetVisor OS offers an optimized path for BFD control packets to reach a destination vRouter. BFD packets are relayed by the NetVisor OS OS from the switch to the destination vRouter using a PCIe link which connects one of the switch ports to the host CPU. The BFD fastpath functionality alleviates BFD timeout/delay issues when other high bandwidth traffic is coming to the host CPU.
The optimized path option can be enabled or disabled using the command:
CLI (network-admin@switch) > system-settings-modify
system-settings-modify |
Use this command to modify system settings. |
Specify one or more of the following options: |
|
optimize-arps|no-optimize-arps |
Enable or disable ARP optimization. |
lldp|no-lldp |
Enable or disable LLDP. |
policy-based-routing|no-policy-based-routing |
Enable or disable Policy-Based Routing (PBR). This enables flexible packet forwarding and routing through user-defined policies. |
optimize-nd|no-optimize-nd |
Enable or disable ND optimization. |
reactivate-mac|no-reactivate-mac |
Enable or disable reactivation of aged out MAC entries. |
reactivate-vxlan-tunnel-mac|no-reactivate-vxlan-tunnel-mac |
Enable or disable reactivation of MAC entries over VXLAN tunnels. |
manage-unknown-unicast|no-manage-unknown-unicast |
Enable or disable unknown unicast management. |
manage-broadcast|no-manage-broadcast |
Enable or disable broadcast management |
block-loops|no-block-loops |
Enable or disable loop blocking. Note: This parameter is available only on NSU, NRU-02, NRU-03, and NRU-S0301 platforms. |
auto-trunk|no-auto-trunk |
Enable or disable auto trunking. |
auto-host-bundle|no-auto-host-bundle |
Enable or disable auto host bundling. |
cluster-active-active-routing|no-cluster-active-active-routing |
Enable or disable active-active routing on a cluster. |
fast-route-download|no-fast-route-download |
Enable or disable fast route download from routesnoop. |
fast-interface-lookup|no-fast-interface-lookup |
Enable or disable fast router interface lookup. |
routing-over-vlags|no-routing-over-vlags |
Enable or disable routing to vLAGs from cluster links. |
source-mac-miss to-cpu|copy-to-cpu |
Specify either of the options as the unknown source MAC learn behavior. |
use-igmp-snoop-l2|use-igmp-snoop-l3 |
Specify whether L2 or L3 tables are to be used for IGMP snooping. |
vle-tracking-timeout <3..30> |
Set a vLE tracking timeout as a value between 3 and 30s. The default timeout is 3s. |
pfc-buffer-limit pfc-buffer-limit-string |
Specify the percent of global system buffer space allowed for PFC. |
cosq-weight-auto|no-cosq-weight-auto |
Specify either of the options to enable or disable automatic weight assignment for CoS (Class of Service) queues based on min-guarantee configuration. |
lossless-mode|no-lossless-mode |
Enable or disable lossless mode. |
stagger-queries|no-stagger-queries |
Stagger igmp/mld snooping queries. |
host-refresh|no-host-refresh |
Enable or disable refreshing host ARP entries to keep L2 entries active. |
proxy-conn-retry|no-proxy-conn-retry |
Enable or disable proxy connection retry. |
proxy-conn-max-retry 0..10 |
Set the maximum number of proxy connection retry attempts as a value between 0 and 10. |
proxy-conn-retry-interval 100..2000 |
Set the number of milliseconds to wait between proxy connection retry attempts. This is a value between 100 and 2000. |
nvos-debug-logging|no-nvos-debug-logging |
Logging mode selection (Direct OR viz. nvlog demon) |
manage-l2-uuc-drop|no-manage-l2-uuc-drop |
Enable or disable L2 UUC (Unknown Unicast Drop) towards data port. |
xcvr-link-debug|no-xcvr-link-debug |
Enable or disable system debug to capture link information. |
fastpath-bfd|no-fastpath-bfd |
Enable or disable BFD fastpath. This feature is disabled by default. |
linkscan-interval 10000..1000000 |
Specify the linkscan interval as a value between 10000μs and 1000000μs. The default value is 150000μs. |
linkscan-mode software|hardware |
Specify the linkscan mode as hardware or software. Software linkscan mode is enabled by default. |
single-pass-flood|no-single-pass-flood |
Enable or disable single-pass flood. |
For example, use the command below to enable BFD fastpath:
CLI (network-admin@switch) > system-settings-modify fastpath-bfd
This command creates a new vFlow which redirects all BFD packets to CPU through the PCIe link.
To see the status of BFD fastpath on a switch, use the command system-settings-show. For example, to view the BFD fastpath status exclusively, use the command:
CLI (network-admin@switch) > system-settings-show format fastpath-bfd
fastpath-bfd: off
NetVisor OS has the transmission queue tx-class-inband-bfd for sending out BFD packets. For reception, the rx-bfd queue is used.
To view the details of the transmission queue, use the command:
CLI (network-admin@switch) > nv-queue-stats-show name tx-class-inband-bfd
To view the receiver queue details, use the command:
CLI (network-admin@switch) > nv-queue-stats-show name rx-bfd