|
OpenStack Pike : Add Compute Nodes
2017/09/01 |
|
Add Compute Nodes to run more instances.
This example is based on the emvironment like follows.
------------+-----------------------------+------------
| |
eth0|10.0.0.30 eth0|10.0.0.51
+-----------+-----------+ +-----------+-----------+
| [ Control Node ] | | [ Compute Node ] |
| | | |
| MariaDB RabbitMQ | | Libvirt |
| Memcached httpd | | Nova Compute |
| Keystone Glance | | |
| Nova API | | |
+-----------------------+ +-----------------------+
|
| [1] |
Install KVM Hypervisor on Compute Node, refer to here.
It's unnecessarry to set Bridge networking on the section [3] of the link. |
| [2] | Install Nova-Compute. |
|
root@node01:~# apt-get -y install nova-compute-kvm
|
| [3] | Configure Nova. |
|
root@node01:~# mv /etc/nova/nova.conf /etc/nova/nova.conf.org
root@node01:~#
vi /etc/nova/nova.conf # create new [DEFAULT] # define own IP address my_ip = 10.0.0.51 state_path = /var/lib/nova enabled_apis = osapi_compute,metadata log_dir = /var/log/nova # RabbitMQ connection info transport_url = rabbit://openstack:password@10.0.0.30 [api] auth_strategy = keystone # enable VNC [vnc] enabled = True vncserver_listen = 0.0.0.0 vncserver_proxyclient_address = $my_ip novncproxy_base_url = http://10.0.0.30:6080/vnc_auto.html # Glance connection info [glance] api_servers = http://10.0.0.30:9292 [oslo_concurrency] lock_path = $state_path/tmp # Keystone auth info [keystone_authtoken] auth_uri = http://10.0.0.30:5000 auth_url = http://10.0.0.30:35357 memcached_servers = 10.0.0.30:11211 auth_type = password project_domain_name = default user_domain_name = default project_name = service username = nova password = servicepassword [placement] auth_url = http://10.0.0.30:35357 os_region_name = RegionOne auth_type = password project_domain_name = default user_domain_name = default project_name = service username = placement password = servicepassword [wsgi] api_paste_config = /etc/nova/api-paste.ini chmod 640 /etc/nova/nova.conf root@node01:~# chgrp nova /etc/nova/nova.conf root@node01:~# systemctl restart nova-compute
|
| [4] | Make sure the status of Nova services on Controle Node like here. If all State is [up], they are running normally. |
|
# discover Compute Nodes root@dlp ~(keystone)# su -s /bin/bash nova -c [nova-manage cell_v2 discover_hosts] root@dlp ~(keystone)# openstack compute service list +----+--------------+--------------+----------+---------+-------+--------------+ | ID | Binary | Host | Zone | Status | State | Updated At | +----+--------------+--------------+----------+---------+-------+--------------+ | 1 | nova- | dlp.srv.worl | internal | enabled | up | 2017-09-01T0 | | | conductor | d | | | | 8:35:29.0000 | | | | | | | | 00 | | 7 | nova- | dlp.srv.worl | internal | enabled | up | 2017-09-01T0 | | | scheduler | d | | | | 8:35:29.0000 | | | | | | | | 00 | | 8 | nova- | dlp.srv.worl | internal | enabled | up | 2017-09-01T0 | | | consoleauth | d | | | | 8:35:28.0000 | | | | | | | | 00 | | 9 | nova-compute | dlp.srv.worl | nova | enabled | up | 2017-09-01T0 | | | | d | | | | 8:35:29.0000 | | | | | | | | 00 | | 10 | nova-compute | node01.srv.w | nova | enabled | up | 2017-09-01T0 | | | | orld | | | | 8:35:29.0000 | | | | | | | | 00 | +----+--------------+--------------+----------+---------+-------+--------------+ |