Setting up multi nodes live migration in Openstack Juno with devstack

Summary

  • Live migration overview

  • Setup Openstack with devstack: one controller node, two compute nodes
  • Configure migrations
  • Live migration test
  • Issue summary

Live migration Overview

Migration enables an administrator to move a virtual-machine instance from one compute host to another. This feature is useful when a compute host requires maintenance. Migration can also be useful to redistribute the load when many VM instances are running on a specific physical machine.

The migration types are:

  • Non-live migration (sometimes referred to simply as 'migration'). The instance is shut down for a period of time to be moved to another hypervisor. In this case, the instance recognizes that it was rebooted.

  • Live migration (or 'true live migration'). Almost no instance downtime. Useful when the instances must be kept running during the migration. The different types of live migration are:

    • Shared storage-based live migration. Both hypervisors have access to shared storage.

    • Block live migration. No shared storage is required. Incompatible with read-only devices such as CD-ROMs and Configuration Drive (config_drive).

    • Volume-backed live migration. Instances are backed by volumes rather than ephemeral disk, no shared storage is required, and migration is supported (currently only available for libvirt-based hypervisors

This page's live migration uses block live migration which does not need shared storage.

Setup Openstack with devstack: one controller node, two compute nodes

Prepare three servers with ubuntu installed. One server is worked as control node, the other two are compute nodes(Note: please named three servers with different hostnames.) and get updates and install git

sudo apt-get update

sudo apt-get install git

Configure controller node

  1. Git devstack Juno release:  git clone https://github.com/openstack-dev/devstack.git -b stable/juno
  2. Add localrc under devstack and add content to it as following

    HOST_IP=your configured IP address
    MULTI_HOST=
    FIXED_RANGE=10.0.0.0/
    FIXED_NETWORK_SIZE=
    ADMIN_PASSWORD=
    MYSQL_PASSWORD=
    RABBIT_PASSWORD=
    SERVICE_PASSWORD=
    SERVICE_TOKEN=
    SCREEN_LOGDIR=/home/stack/log/screen
    LOGFILE=/home/stack/log/stack.log
    LOGDAYS=
    PUBLIC_INTERFACE=eth0 #actual name of your network interface
    FLAT_INTERFACE=eth0
  3. Run devstack to install openstack: ./stack.sh

Configure compute node

  1. Git devstack Juno release:  git clone https://github.com/openstack-dev/devstack.git -b stable/juno
  2. Add localrc under devstack and add content as done in controller node. and also add following to it.

    SERVICE_HOST=your controller ip
    
    HOST_IP=host ip
    
    MULTI_HOST=
    
     FIXED_RANGE=10.0.0.0/
    
    FIXED_NETWORK_SIZE=
    
     ADMIN_PASSWORD=
    
    MYSQL_PASSWORD=
    
    RABBIT_PASSWORD=
    
    SERVICE_PASSWORD=
    
    SERVICE_TOKEN=
    
    SCREEN_LOGDIR=/home/stack/log/screen
    
    LOGFILE=/home/stack/log/stack.log
    
    LOGDAYS=
    
    #PUBLIC_INTERFACE=eth0
    
    FLAT_INTERFACE=eth0
    
    DATABASE_TYPE=mysql
    
    Q_HOST=$SERVICE_HOST
    
    MYSQL_HOST=$SERVICE_HOST
    
    RABBIT_HOST=$SERVICE_HOST
    
    GLANCE_HOSTPORT=$SERVICE_HOST:
    
    KEYSTONE_AUTH_HOST=$SERVICE_HOST
    
    KEYSTONE_SERVICE_HOST=$SERVICE_HOST
    
    ENABLED_SERVICES=n-cpu,n-net,n-api,c-sch,c-api,c-vol
  3. Add some variables which would use during install openstack
    Copy following info to a shell file:

    #!/bin/bash
    
    export OS_AUTH_URL=http://controller'ip:5000/v2.0
    export OS_TENANT_ID=69eaaebc44174418956d4dd104f2be76
    export OS_TENANT_NAME="admin" export OS_USERNAME="admin"
    export OS_PASSWORD=
    export OS_VOLUME_API_VERSION=

    Update the OS_AUTH_URL with your controller node ip. And use command:mysql -uroot -p123456 -e "select * from keystone.project;"  to get OS_TENANT_ID and update it, then run this shell

  4. Run devstack to install openstack: ./stack.sh

Configure migrations

  1. Host name configuration
    Add three server's host names to /etc/hosts to be sure that every node can ping successful to others

    server's IP   server's host name
  2. Configure /etc/sysconfig/iptables file to allow libvirt listen on TCP port 16509 and add a record accepting KVM communication on TCP port within the range from 49152 to 49261
    Note: ubuntu does not have sysconfig folder under etc, need create it manually.

    -A INPUT -p tcp -m multiport --ports  -m comment --comment "libvirt" -j ACCEPT
    
    -A INPUT -p tcp -m multiport --ports : -m comment --comment "migration" -j ACCEPT
  3. Configuration libvirt
    Enable libvirt listen flag at /etc/sysconfig/libvirtd file: add

    LIBVIRTD_ARGS="-listen"

    Configure /etc/libvirt/libvirtd.conf:

    listen_tls =
    listen_tcp =
    auth_tcp = "none"

    Configure /etc/init/libvirt-bin.conf, modify exec /usr/sbin/libvirtd  to exec /usr/sbin/libvirtd -l
    Configure /etc/libvirt/qemu.conf,  modify security_driver="none"

    Restart libvirt, after executing the command, ensure that libvirt is successfully restarted: sudostop libvirt-bin && sudo start libvirt-bin

    ps -ef |grep libvirt

  4. Nova configuration
    Modify:force_config_drive = None(value from always to None)

    Add:live_migration_flag=VIR_MIGRATE_UNDEFINE_SOURCE,VIR_MIGRATE_PEER2PEER,VIR_MIGRATE_LIVE  (under default)

  5. Restart service on each node
    screen -s stack -X quit
    ./rejoin-stack.sh

Live migration test

Before start live migration test, to be sure that the nova compute service are ready

      

  1. create instance and attach volume
    nova boot --flavor=flavor_id --image=image_id instance_name
    nova volume-attach vm_id volume_id auto

    +----------+--------------------------------------+
    | Property | Value |
    +----------+--------------------------------------+
    | device | /dev/vdb |
    | id | 3deb171f-8d61-4df9-8e6e-29d86575e78e |
    | serverId | bb12f3b0-c17e-49f7-b482-255258931eb4 |
    | volumeId | 3deb171f-8d61-4df9-8e6e-29d86575e78e |
    +----------+--------------------------------------+
  2. How to mount the volume which attached to the instance
    1> execute command to ssh to the instance, for example: ssh cirros@10.0.0.2
    2> execute command to list the partition tables for the installed devices: sudo fdisk -l
         see:Disk /dev/vdb......
    3> create a filesystem on the  device: sudo mkfs.ext4 /dev/vdb
    4> create a dictionary and mount the volume:
         sudo mkdir /data
         sudo mount /dev/vdb /data/
    5> then we can create folder or file on it
  3. nova show instance_id
  4. nova live-migration --block_migrate vm_id target_server_hostname
  5. check the host before and after migration

Issue summary

1. n-api could not start when install openstack with devstack
Solution: Restart controller node and rejoin the service, then run unstack and stack on compute node

2. Get error:libvirtError: operation failed: Failed to connect to remote libvirt URI qemu+tcp://computer247/system: Unable to resolve address 'computer247' service '16509': Name or service not known
Solution: check the iptables in configure migration step 2.

3. Live Migration failure: Invalid value '4-7,12-15' for 'cpuset.cpus': Invalid argument
Solution: Check the CPU info in your compute nodes, if there have different architecture of cpus, it would be fail to migrate

4. Compute node's nova-cpu service could not enable:


Solution: Check the libvirt settings in configure migration section, and restart libvirt-bin service, then run command in both compute node and controller node:
screen -s stack -X quit
./rejoin-stack.sh

5. Get error:


Solution: It's about nova configuration, check /etc/nova/nova.conf:
Modify:force_config_drive = None(value from always to None)      Add:live_migration_flag=VIR_MIGRATE_UNDEFINE_SOURCE,VIR_MIGRATE_PEER2PEER,VIR_MIGRATE_LIVE  (under default)
then rejoin stack in compute node

6. Getting error:screen-n-cpu.2015-06-02-184950.log:2015-06-04
00:16:45.797 ERROR nova.virt.libvirt.driver [-] [instance:
d1b56987-4691-4259-897f-ba4ce3e71260] Live Migration failure: Failed to
open file '/dev/disk/by-path/ip-9.115.246.45:3260-iscsi-iqn.1986-03.com.ibm:2145.v7k41.node1-lun-4': No such file or directory
Solution:  No solution currently by me, it is a bug and would be fixed

7. Gettting
error when rejoin stack:2015-06-09 14:02:33.154 TRACE
cinder.openstack.common.threadgroup ArgumentError: Could not parse
rfc1738 URL from string ''"


Solution:
modify the node sql_connection  in /etc/cinder/cinder.conf and
/etc/nova/nova.conf with( It has no value as default): take following as
example
/etc/cinder/cinder.conf:  sql_connection = mysql://root:123456@9.115.246.185/cinder?charset=utf8
/etc/nova/nova.conf:   sql_connection = mysql://root:123456@9.115.246.185/nova?charset=utf8

Setting up multi nodes live migration in Openstack Juno with devstack的更多相关文章

  1. Deploying Cloud Foundry on OpenStack Juno and XenServer (Part I)

    link http://rabbitstack.github.io/deploying-cloud-foundry-on-openstack-juno-and-xenserver-part-i/ Cl ...

  2. 使用openshit在ubuntu14.04下一键部署openstack(juno版本)

    一.基本介绍 本实验是在vmware workstation上虚拟机ubuntu14.04(64bit,desktop)上部署openstack(Juno版本).采用的工具是openshit.open ...

  3. openStack juno for ubuntu12-04

    <一,preinstall basic conf,pre Env> 1,pwgen(openssl rand -hex 10) some Open-Stack services add a ...

  4. OpenStack Juno 版本发布——支持Spark和NFV[转]

    作者:郑晨,OpenStack中国社区,转载请注明出处 美国时间2014年10月16日,OpenStack Juno版本正式发布,这是OpenStack开源云计算项目自2010年创立以来的第10个版本 ...

  5. 一键安装openstack juno 之controller node.

    原文名称: OpenStack Juno Scripted Installation on CentOS 7 Step I:  本机信息配置 CONTROLLER_IP=192.168.173.133 ...

  6. [openStack]使用Fuel安装OpenStack juno的fuel_master

    安装OpenStack是一件很复杂的事情,特别是在想目中,如果一个组件一个组件,一台一台的coding部署,估计太消耗时间,而且出错的概率很高,所以使用工具推送部署的效率就很高了,而且必须得可靠.mi ...

  7. vmware vms migration to openstack

    Converting a VMware Workstation virtual machine to KVM Leave a commentPosted by rbgeek on August 13, ...

  8. VMware migration to openstack kvm

  9. Centos7 install Openstack Juno (RDO) (转载)

    原文地址:http://www.hdume.com/centos-7-0%E5%AE%89%E8%A3%85openstack/ 1.安装系统,Centos7镜像采用CentOS-7.0-1406-x ...

随机推荐

  1. hive 日志配置/表头配置

    1.日志配置,拷贝hive/conf下的hive-log4j2.properties.template为hive-log4j2.properties,修改日志目录,接下来在创建hive/logs,目录 ...

  2. experiment : 在私有堆和默认进程堆中, 测试能分配的堆空间总和, 每次能分配的最大堆空间

    实验环境: Win7X64Sp1 + vs2008,  物理内存16GB. 实验结论: *  进程堆的最大Size并没有使用完剩余的物理内存    *  每次能分配的最大堆空间接近2M, 不管是私有堆 ...

  3. MySQL实现类似Oracle中的nextval和currval

    CREATE TABLE `sequence` ( `seq_name` varchar(50) NOT NULL, `current_val` int(11) NOT NULL, `incremen ...

  4. nditer —— numpy.ndarray 多维数组的迭代

    1. Single array iteration >>> a = np.arange(6).reshape(2,3) >>> for x in np.nditer ...

  5. 编译freetype 的dll

    因需要给python使用freetype库,so需要一个freetype的dll 2 steps 1. 在VC中设置输出为动态链接库 2. 修改ftoption.h 在284行增加2行代码即可 /** ...

  6. echarts的一些基础笔记

    图表标题 title: { x: "left", // 水平安放位置,默认为左对齐,可选为: // 'center' ¦ 'left' ¦ 'right' // ¦ {number ...

  7. 从编译,执行过程理解c#

    上节我们说过C#所开发的程序源代码并不是编译成能够直接在操作系统上执行的二进制代码.与Java类似,它被编译成为中间代码,然后通过.NET Framework的虚拟机——被称之为通用语言运行时(CLR ...

  8. Java中文件的上传与下载

    文件的上传与下载主要用到两种方法: 1.方法一:commons-fileupload.jar  commons-io.jar apache的commons-fileupload实现文件上传,下载 [u ...

  9. 监听WPF依赖属性

    原文:监听WPF依赖属性 当我们使用依赖属性的时候,有时需要监听它的变化,这在写自定义控件的时候十分有用, 下面介绍一种简单的方法.   如下使用DependencyPropertyDescripto ...

  10. Leetcode 169 Majority Element 分治

    在一个长度为n的数组中找出出现次数超过(n+1)/2次的数 说明请参考编程之美中的2.3 class Solution { public: int majorityElement(vector< ...