On earlier versions (as of now 5.15.x is still supported), the preferred way is the manage_ovs script. Basically, it is the scripted way of configuring the bond via ovs-vsctl which runs on CVMs and reduces the possibility of manual mistakes.
We usually don’t recommend using ovs-vsctl commands and they can be used mostly for troubleshooting purposes.
Isn’t it just that manage_ovs is done from inside the CVM, but ovs_vsctl is done from the host?
To configure the LACP on my home CE setup, I did it via ovs_vsctl on the hosts:
ovs-vsctl set port br0-up other_config:lacp-fallback-ab=true ovs-vsctl set port br0-up bond_mode=balance-tcp ovs-vsctl set port br0-up lacp=active
I seem to remember I got cut off after running one of these commands (I think it was the second one) and had to run the last command from the keyboard plugged into the host (or ILO/IPMI) :)
Manage_ovs is just a “script” that configures host networking with the recommended settings and it’s launched by the CVM (CVM communicates internally with the AHV host via the private IP 192.168.5.1) it seems to be the “recommended” way to configure host networking whereas ovs-vsctl and ovs-appctl are launched directly from the AHV host and are used to adjust specific parameters like bond-rebalance-interval on balance-slb for example
I always use manage_ovs to configure AHV Networking on production systems and ovs-vsctl only to set up AHV VLAN with ovs-appct to check LACP negotiation.
Manage_ovs is just a “script” that configures host networking with the recommended settings and it’s launched by the CVM (CVM communicates internally with the AHV host via the private IP 192.168.5.1) it seems to be the “recommended” way to configure host networking whereas ovs-vsctl and ovs-appctl are launched directly from the AHV host and are used to adjust specific parameters like bond-rebalance-interval on balance-slb for example
I always use manage_ovs to configure AHV Networking on production systems and ovs-vsctl only to set up AHV VLAN with ovs-appct to check LACP negotiation.
Isn’t it just that manage_ovs is done from inside the CVM, but ovs_vsctl is done from the host?
To configure the LACP on my home CE setup, I did it via ovs_vsctl on the hosts:
ovs-vsctl set port br0-up other_config:lacp-fallback-ab=true ovs-vsctl set port br0-up bond_mode=balance-tcp ovs-vsctl set port br0-up lacp=active
I seem to remember I got cut off after running one of these commands (I think it was the second one) and had to run the last command from the keyboard plugged into the host (or ILO/IPMI) :)
Cheers,
Steve
Thank you Steve but I’ve get confused because in nutanix portal I’ve seen an article about manage_ovs is recommended way in aos version AOS >= 5.19 and < 5.20.2 my AOS version is 5.15.4
On earlier versions (as of now 5.15.x is still supported), the preferred way is the manage_ovs script. Basically, it is the scripted way of configuring the bond via ovs-vsctl which runs on CVMs and reduces the possibility of manual mistakes.
We usually don’t recommend using ovs-vsctl commands and they can be used mostly for troubleshooting purposes.