One OpenStack instance is mapped to an EFA tenant. This section describes the steps involved in setting up the tenant on EFA.
| Steps | Description |
|---|---|
| Create a Tenant | Create an EFA tenant. The name of the tenant should be the same as the region_name in Neutron. |
| Create Lags/Port Channel | Create lag according to the physical topology. The names of the lags must be used in defining the topology mapping on Neutron. |
| Create Shared Tenant | Create an EFA tenant that contains the shared resources across multiple EFA tenants. The name of the tenant must be used as region_sharedin ML2 configurations. |
| EFA REST TOKEN | Used as API key efa_rest_token in ML2 configurations for secure connection towards EFA. |
The following are the EFA CLI commands used to achieve the steps above:
| EFA CLI Command | |
|---|---|
efa tenant create --name VIM_1 --vlan-range 2-4090 --l2-vni-range
1-5000 --port 10.24.14.133.[0/1], 10.24.14.134.[0/1],10.24.14.135.[0/1],
10.24.14.136.[0/1-2] |
Create Tenant VIM_1 |
efa tenant po create --name lag_1 --tenant VIM_1 --port
10.24.14.133.[0/1], 10.24.14.134.[0/1] --speed 10Gbps --negotiation active
|
Lag on C1 |
efa tenant po create --name lag_2 --tenant VIM_1 --port
10.24.14.135.[0/1], 10.24.14.136.[0/1] --speed 10Gbps --negotiation active
|
Lag on C2 |
efa tenant create --name SHARED_TENANT --port 10.24.14.191.[0/1],
10.24.14.192.[0/1] –type shared
|
Shared tenant on border leaf |
efa tenant po create --name lag_4 --tenant SHARED_TENANT --port
10.24.14.133.[0/1], 10.24.14.134.[0/1] --speed 10Gbps --negotiation active
|
Border leaf lag created on shared tenant |
efa auth client register --name VIM_1 --type openstack
|
Register VIM_1 tenant for OpenStack access |
efa auth apikey generate --client-id <client_id> |
Use the Client ID generated in client register for API key generation
Use the generated API Key as a the efa_rest_token in OpenStack |