标签：
陈沙克在他的经典文章里分析了多个安装部署工具，用于openstack部署。
当然时代变了，所以很多这些工具都改头换面了。
目前看，
1 Mirantis Fuel仍然是最好的商业用选择，虽然还有瑕疵。
fuel在7和7之后已经提供开源的plugin插件，例如contrail的插件。并将fuel plugin方案放入openstack发展会中。
fuel在6的时候已经在提供商业化的plugin了。
fuel目前的8有个讨厌的地方就是需要联网取些内容（是我网的问题，联网去取docker image供slave用），对centos7.0的版本兼容性也不好。（是centos7.0的问题，其升kernel到7.1很大可能会挂）
fuel的瑕疵主要是有些场景会莫名的出现异常，部分删除和增加云会导致异常，一些硬盘大小识别在安装快结束时才告诉你容量不足。（主要是6的问题，8因网络问题至今无感）
2 RDO，即原来的packstack
RDO无光盘，也是出了名的网络安装，没网怎么活？
RDO安装的粒度必须是全openstack
RDO安装了ceilemeter，第二次换了ip（如果是dhcp申请的），重装，如果不选择ceilemeter，则原来的ceil不会自动消亡，还在那里盲目发信息要求通讯。
3 其他puppet类，如陈所言。
其中rackspace，我没找到光盘，也没找到脚本。rackspace是不是已经不怎么玩openstack了？
4 ansible和kolla
ansible比较方便，因为总有人讨厌puppet，在下不懂puppet，所以也不愿意碰
kolla是ansible语言的上层开发。提供了以openstack 服务级粒度的openstack安装部署。
kolla可以支持对失败的openstack的nova，keystone服务重新单独部署。
kolla还基于docker
fuel7.0安装，和fuel5.0、6安装相似，这次小弟是在virtualbox上安装的。
小弟修改了config.sh
主要改动点
1 master， 1compute, 1controller,
给计算节点多点资源。
不过安装成功后，惊讶的发现fuel7很智能的把整张盘的大小都给了compute，controller
内存上各节点的配置则和config.sh相同。
贴上config.sh
#!/bin/bash
# Copyright 2013 Mirantis, Inc.
#
# Licensed under the Apache License, Version 2.0 (the "License"); you may
# not use this file except in compliance with the License. You may obtain
# a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
# WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the
# License for the specific language governing permissions and limitations
# under the License.
source ./functions/memory.sh
# Get the first available ISO from the directory 'iso'
iso_path=`ls -1t iso/*.iso 2>/dev/null | head -1`
# Every Mirantis OpenStack machine name will start from this prefix
vm_name_prefix=fuel-
# By default, all available network interfaces vboxnet won't be removed,
# if their IP addresses don't match with fuel_master_ips (10.20.0.1 172.16.0.254
# 172.16.1.1)
# If you want to remove all existing vbox interfaces, then use rm_network=1
# 0 - don't remove all vbox networks. Remove only fuel networks if they exist
# 1 - remove all vbox networks
rm_network=0
# Please add the IPs accordingly if you going to create non-default NICs number
# 10.20.0.1/24 - Mirantis OpenStack Admin network
# 172.16.0.1/24 - OpenStack Public/External/Floating network
# 172.16.1.1/24 - OpenStack Fixed/Internal/Private network
# 192.168.0.1/24 - OpenStack Management network
# 192.168.1.1/24 - OpenStack Storage network (for Ceph, Swift etc)
fuel_master_ips="10.20.0.1 172.16.0.254 172.16.1.1"
# Network mask for fuel interfaces
mask="255.255.255.0"
# Determining the type of operating system and adding CPU core to the master node
case "$(uname)" in
Linux)
os_type="linux"
if [ "$(nproc)" -gt "1" ]; then
vm_master_cpu_cores=2
else
vm_master_cpu_cores=1
fi
;;
Darwin)
os_type="darwin"
mac_nproc=`sysctl -a | grep machdep.cpu.thread_count | sed 's/^machdep.cpu.thread_count\:[ \t]*//'`
if [ "$mac_nproc" -gt "1" ]; then
vm_master_cpu_cores=2
else
vm_master_cpu_cores=1
fi
;;
CYGWIN*)
os_type="cygwin"
if [ "$(nproc)" -gt "1" ]; then
vm_master_cpu_cores=2
else
vm_master_cpu_cores=1
fi
;;
*)
echo "$(uname) is not supported operating system."
exit 1
;;
esac
# Master node settings
vm_master_memory_mb=1536
vm_master_disk_mb=65535
# Master node access to the internet through the host system, using VirtualBox NAT adapter
vm_master_nat_network=192.168.200.0/24
vm_master_nat_gateway=192.168.200.2
# These settings will be used to check if master node has installed or not.
# If you modify networking params for master node during the boot time
# (i.e. if you pressed Tab in a boot loader and modified params),
# make sure that these values reflect that change.
vm_master_ip=10.20.0.2
vm_master_username=root
vm_master_password=r00tme
vm_master_prompt='root@fuel ~]#'
# The number of nodes for installing OpenStack on
# - for minimal non-HA installation, specify 2 (1 controller + 1 compute)
# - for minimal non-HA with Cinder installation, specify 3 (1 ctrl + 1 compute + 1 cinder)
# - for minimal HA installation, specify 4 (3 controllers + 1 compute)
if [ "$CONFIG_FOR" = "16GB" ]; then
cluster_size=5
elif [ "$CONFIG_FOR" = "8GB" ]; then
cluster_size=3
else
# Section for custom configuration
cluster_size=2
fi
# Slave node settings. This section allows you to define CPU count for each slave node.
# You can specify CPU count for your nodes as you wish, but keep in mind resources of your machine.
# If you don't, then will be used default parameter
if [ "$CONFIG_FOR" = "16GB" ]; then
vm_slave_cpu_default=1
vm_slave_cpu[1]=1
vm_slave_cpu[2]=1
vm_slave_cpu[3]=1
vm_slave_cpu[4]=1
vm_slave_cpu[5]=1
elif [ "$CONFIG_FOR" = "8GB" ]; then
vm_slave_cpu_default=1
vm_slave_cpu[1]=1
vm_slave_cpu[2]=1
vm_slave_cpu[3]=1
else
# Section for custom configuration
vm_slave_cpu_default=1
vm_slave_cpu[1]=4
vm_slave_cpu[2]=1
fi
# This section allows you to define RAM size in MB for each slave node.
# Keep in mind that PXE boot might not work correctly with values lower than 768.
# You can specify memory size for the specific slaves, other will get default vm_slave_memory_default
# Mirantis OpenStack 3.2 controllers require 1280 MiB of RAM as absolute minimum due to Heat!
# You may comment out all the following memory parameters to use default value for each node.
# It is recommended if you going to try HA configurations.
# for controller node at least 1.5Gb is required if you also run Ceph and Heat on it
# and for Ubuntu controller we need 2Gb of ram
# For compute node 1GB is recommended, otherwise VM instances in OpenStack may not boot
# For dedicated Cinder, 768Mb is OK, but Ceph needs 1Gb minimum
if [ "$CONFIG_FOR" = "16GB" ]; then
vm_slave_memory_default=1536
vm_slave_memory_mb[1]=2048
vm_slave_memory_mb[2]=2048
vm_slave_memory_mb[3]=2048
vm_slave_memory_mb[4]=2048
vm_slave_memory_mb[5]=2048
elif [ "$CONFIG_FOR" = "8GB" ]; then
vm_slave_memory_default=1024
vm_slave_memory_mb[1]=1536
vm_slave_memory_mb[2]=1536
vm_slave_memory_mb[3]=1536
else
# Section for custom configuration
vm_slave_memory_default=1024
vm_slave_memory_mb[1]=12048
vm_slave_memory_mb[2]=1024
vm_slave_memory_mb[3]=1024
fi
# Within demo cluster created by this script, all slaves (controller
# and compute nodes) will have identical disk configuration. Each
# slave will have three disks with sizes defined by the variables below. In a disk configuration
# dialog you will be able to allocate the whole disk or it's part for
# operating system (Base OS), VMs (Virtual Storage), Ceph or other function,
# depending on the roles applied to the server.
# Nodes with combined roles may require more disk space.
vm_slave_first_disk_mb=85535
vm_slave_second_disk_mb=35535
vm_slave_third_disk_mb=65535
# Set to 1 to run VirtualBox in headless mode
headless=0
skipfuelmenu="no"
标签：
原文地址：http://blog.csdn.net/anzhuangguai/article/details/51334362
