The sections below
provide the information on installation of Nexus Fabric Enabler.
Install Nexus
Fabric Enabler
Ensure your
OpenStack installation is running correctly based on required setting discussed
in the above sections.
On the control node in the
OpenStack installation, get the source file
git clone -b enabler_1_0_0
https://github.com/CiscoSystems/fabric_enabler
openstack_fabric_enabler
Go to the enabler directory
created in the above step
cd
~/openstack_fabric_enabler
modify enabler_conf.ini
Following are the
sections for modification in dcnm-mgmt interface (WebGui interface, Rest API
and so on):
[dcnm]
dcnm_ip = 1.1.1.1
dcnm_user = username
dcnm_amqp_user = amqpdcnmusername
dcnm_password = password
gateway_mac = 20:20:00:00:00:AA
segmentation_id_min = segidmin
segmentation_id_max = segidmax
Ensure to replace
the values dcnm_ip, username, amqpdcnmusername, password, segidmin and segidmax
based on your setup. The enhanced-fabric-mgmt (ssh, LDAP and so on) command is
introduced to solve the problem described in the vswitch mobility scenario. In
the absence of this config, there will be greater traffic loss in the scenarios
where a host move occurs for the hosts that were connected through vswitch.
Note |
The user has
to ensure that the segment ID should not be overlapping with other segment ID
ranges (for example the Cisco Prime DCNM segment ID is using, default 30,000 to
49,999).
|
This is the
section about Cisco Prime DCNM, which you have installed separately and set the
right access credentials to be used by OpenStack Nexus Fabric Enabler. Ensure
the gateway_mac matches your PoAP template setting in Cisco Prime DCNM for your
leaf switch and you use the right range of segment IDs administrated by your
fabric manager.
[dfa_mysql]
# MYSQL DB connection parameter
connection = mysql://username:password@localhost/cisco_dfa?charset=utf8
Ensure to replace
the username and password based on your setting.
[dfa_rpc]
# Transport URL parameter for RPC
transport_url = 'rabbit://username:password@%(ip)s:5672//'
Ensure to replace
the username and password based on your setting.
[general]
compute_user = username
compute_passwd = password
Ensure to replace
the username and password based on your setting.
Checking conf files
The
keystone_authtoken in neutron.conf and nova.conf should be similar to the
following setting:
For neutron.conf:
[keystone_authtoken]
signing_dir = /var/cache/neutron
auth_uri = http://<ip address of controller>:5000/v2.0
cafile = /opt/stack/data/ca-bundle.pem
identity_uri = http://<ip address of controller>:35357
auth_host = <ip address of controller>
auth_port = 35357
auth_protocol = http
admin_tenant_name = service
admin_user = neutron
admin_password = password
For nova.conf:
[keystone_authtoken]
signing_dir = /var/cache/nova
admin_password = password
admin_user = nova
admin_tenant_name = service
auth_uri = http://<ip address of controller>:5000/v2.0
cafile = /opt/stack/data/ca-bundle.pem
identity_uri = http://<ip address of controller>:35357
auth_protocol = http
auth_port = 35357
auth_host = <ip address of controller>
Note |
It requires all
control/compute nodes that have same username/password and this is your Linux
account on the servers to run as control/compute nodes.
|
Network Time Protocol
(NTP)
The NTP
installation is required by openstack (refer to openstack installation guide
for NTP installation requirements).
In case of clock
drift in openstack node, the following command can be executed to sync the node
manually:
On ubuntu machine:
sudo service ntp stop
sudo ntpdate -s <your ntp server>
sudo service ntp start
On RedHat:
$ sudo vi /etc/sysconfig/ntpd
OPTIONS="-u ntp:ntp -p /var/run/ntpd.pid"
OPTION="-x -u ntp:ntp -p /var/run/ntpd.pid"
$ sudo service ntpd restart
Run python
setup_enabler
Using the source
file downloaded from
git clone
https://github.com/CiscoSystems/fabric_enabler
openstack_fabric_enabler, the script will install all the necessary
modules on both the control and compute nodes/servers and then bring them up.