Extreme Networks OpenStack Plugin 2.0 Installation Guide

Copyright © 2014 Extreme Networks AccessAdapt, Alpine, Altitude, BlackDiamond, Direct Attach, EPICenter, ExtremeWorks Essentials, Ethernet Everywhere, Extreme Enabled, Extreme Ethernet Everywhere, Extreme Networks, Extreme Standby Router Protocol, Extreme Turbodrive, Extreme Velocity, ExtremeWare, ExtremeWorks, ExtremeXOS, Go Purple Extreme Solution, ExtremeXOS ScreenPlay, ReachNXT, Ridgeline, Sentriant, ServiceWatch, Summit, SummitStack, Triumph, Unified Access Architecture, Unified Access RF Manager, UniStack, XNV, the Extreme Networks logo, the Alpine logo, the BlackDiamond logo, the Extreme Turbodrive logo, the Summit logos, and the Powered by ExtremeXOS logo are trademarks or registered trademarks of Extreme Networks, Inc. or its subsidiaries in the United States and/or other countries. sFlow is the property of InMon Corporation. iBooks is property of Apple, Inc. Specifications are subject to change without notice. All other registered trademarks, trademarks, and service marks are property of their respective owners. For additional information on Extreme Networks trademarks, please see: http:// www.extremenetworks.com/company/legal/trademarks/. 120873-00 Rev 2

Table of Contents Chapter 1: Prerequisites 4 Software Requirements 4 Hardware Requirements 4 OpenStack Requirements 5

Chapter 2: Reference Topology 6 Reference Topology Setup 6

Chapter 3: Installing Ubuntu 12.04 LTS on the Hosts 10 Installing Ubuntu 12.04 LTS on the Hosts 10

Chapter 4: Preparing the Servers and Switches 12 Updating the Servers 12 Setting up the Ethernet Port for the Management Interface 12

Chapter 5: Installing OpenStack Plugin 2.0 on the Servers 13 Downloading and Preparing OpenStack 13 Installing OpenStack on the OSController 13 Installing OpenStack on the OSHosts 16

Chapter 6: Configuring Extreme Networks Switches 19 Configuring Control and TOR Switches 19 Configuring the TOR1 Switch 19 Configuring the TOR2 Switch 20 Configuring the Control 1 Switch 20 Configuring the Control 2 Switch 20

Chapter 7: Starting and Stopping Extreme Networks OpenStack 22 Starting and Stopping Extreme Networks OpenStack 22 Modifying the Devstack localrc Parameters 23 Starting OpenStack on OSController 24 Starting OpenStack on OSHost1 and OSHost2 25 Verifying the OSController and OSHosts 25 Populating the Topology Database 26 Configuring the Network Fabric—LAG/MLAG 31 Shutting Down OpenStack on All Servers 33 Logs 33

Chapter 8: Managing Tenants and Virtual Machines 35 Creating Tenants 35 Creating Tenants Using Python Script and Configuration File (L3 Agent) 35 Creating Tenants Using Python Script and Configuration File (Virtual Routers) 37 Verifying TOR Switch Configuration after Tenant Creation (L3 Agent) 40 Verifying TOR Switch Configuration after Tenant Creation (Virtual Routers) 42 Creating Tenant Virtual Machine Instances 44 Migrating Tenant Virtual Machine Instances (Live Migration) 50 Deleting Tenant Virtual Machine Instances 51

Chapter 9: Resetting the Testbed 53 Appendix A: Glossary 54

Extreme Networks OpenStack Plugin 2.0 Installation Guide

3

1 Prerequisites Software Requirements Hardware Requirements OpenStack Requirements This chapter explains the prerequisites for installing the Extreme Networks OpenStack Plugin 2.0.

Software Requirements You need the following software installed:

• • •

Extreme Networks' ExtremeXOS operating system release 15.3.2 or 15.3.3 ( www.extremenetworks.com/products/extreme-xos.aspx). Extreme Networks OpenStack Plugin 2.0 software package (request download package from Extreme Networks). Ubuntu 12.04 LTS (Precise) image with KVM (www.ubuntu.com/download).

Hardware Requirements You need the following hardware if you want to install the reference topology setup (see Reference Topology Setup on page 6):

• • •

OSController (OpenStack Cloud Controller with Quantum Server and Network Host; main server)— 64-bit x86 processor, 8GB RAM (minimum), 7 NICs OSHost1, OSHost2 (OpenStack compute hosts; hosts VMs only)—64-bit x86 processor, 8GB RAM (minimum), 7 NICs TOR1, TOR2—Extreme Networks switch (recommended: Summit X460, X480, or X670)

We support any switch running ExtremeXOS release 15.3.2 or 15.3.3: Stackable Switches

• • •

Summit X670 (www.extremenetworks.com/product/summit-x670-series) Summit X480 (http://www.extremenetworks.com/product/summit-x480-series) Summit X460 (http://www.extremenetworks.com/product/summit-x460-series)

Chassis-Based Switches

• •

BlackDiamond X8 (http://www.extremenetworks.com/product/blackdiamond-x-series) BlackDiamond 8800 (http://www.extremenetworks.com/product/blackdiamond-8800-series)

Prerequisites

OpenStack Requirements OpenStack requirements are available at: http://docs.openstack.org/grizzly/openstack-ops/content/index.html.

Extreme Networks OpenStack Plugin 2.0 Installation Guide

5

2 Reference Topology Reference Topology Setup

Reference Topology Setup The reference topology setup consists of three servers: one controller (OSController) and two compute nodes (OSHost1 and OSHost2); two control switches (CTRL1 and CTRL2); and two "top of rack" switches (TOR1 and TOR2). This setup uses redundancy where possible; for the servers this means that bonding is used to connect to the TOR switches, as well as the control switches. There are distinct networks for data, storage, control, and management—each using its own set of NICs, or in the case of the management port, a single NIC. The control network is not the management network. In the following setup, OpenStack services are behind the control network IP.

Figure 1: Complete Reference Topology Setup

Reference Topology

Control network (in red) is used for exchanging control messages between the OpenStack components (servers) and ExtremeXOS switches. Management network (in blue) is used for out-of-band access. It requires Internet access during installation, but can run through a SNAT layer. It can use any routable subnet. Data network (in purple) is used for data traffic from the tenant VMs on the servers. It is also used for DHCP request/reply between tenant VMs and per-tenant DHCP/NAT server on the controller. It is also used for tenant data traffic to/from public/external network through per-tenant gateway on the controller. Storage network (in green) is used for storage traffic from the tenant VMs on the servers.

Figure 2: Complete Reference Topology Setup with Border Gateways The following four figures show in isolation each of the logical networks (data, storage, control, and management) within the reference topology.

Extreme Networks OpenStack Plugin 2.0 Installation Guide

7

Reference Topology

Figure 3: Control Network

Figure 4: Data Network

Extreme Networks OpenStack Plugin 2.0 Installation Guide

8

Reference Topology

Figure 5: Storage Network

Extreme Networks OpenStack Plugin 2.0 Installation Guide

9

3 Installing Ubuntu 12.04 LTS on the Hosts

Installing Ubuntu 12.04 LTS on the Hosts

Installing Ubuntu 12.04 LTS on the Hosts The Extreme Networks OpenStack software package is based on DevStack (http://devstack.org), which assumes a specific version of Ubuntu—12.04 LTS (Precise) for this release. To download and install the Ubuntu Server on the hosts (controller and compute nodes): 1

Go to www.ubuntu.com/download/server and download Ubuntu Server 12.04.3 LTS 64-bit (ubuntu-12.04.3-server-amd64.iso). For specific instructions, see the Ubuntu Installation Guide.

2 Load the image on a CD-ROM or USB memory stick. 3 Boot the system from the CD-ROM or USB memory stick. 4 On the Language screen, select the language for the install. 5 On the Ubuntu screen, select Install Ubuntu Server. 6 On the Select a language screen, select the language for the installation text and for the installed program. 7 On the Select your location screen, select your country. 8 On the Configure the keyboard screen, select:

• •

Yes—Installation program attempts to detect your keyboard layout. Go to Step 9 on page 10. No—Choose your keyboard layout from a list:

a Select the country of origin for the keyboard from the list. b Select the layout for the keyboard from the list. 9 On the Configure the network screen, enter a hostname (for example, OSController, OSHost1, or OSHost2). Important It is recommended that you use the naming convention "OSController", "OSHost1", and "OSHost2" to make it easier to follow the rest of the procedures in this Installation Guide. However, you may use whatever naming convention you prefer, but you must be consistent. OpenStack uses hostnames for communication, and several files (hosts files, configuration files, etc.) depend on consistent use of hostnames. 10 On the Set up users and passwords screen, type:

• • •

stack for the user stack for the username stack for the password

Installing Ubuntu 12.04 LTS on the Hosts

11 On the Set up users and passwords screen, re-type the user password to verify it. 12 On the Set up users and passwords screen, select yes when you are warned that the password consists of less than eight characters. 13 On the Set up users and passwords screen, select no when prompted to encrypt your home directory. 14 On the Configure the clock screen, select your time zone. 15 On the Partition disks screen, select Guided - user entire disk and set up LVM. 16 On the Partition disks screen, select the disk to partition. 17 On the Partition disks screen, select yes to confirm that you want to write the changes to disk and configure LVM. 18 On the Partition disks screen, select Continue to accept the maximum amount for the volume group for guided partitioning (21.2 GB). 19 On the Partition disks screen, select yes to accept writing the changes to disk. 20 On the Configure the package manager screen, type a proxy path, if needed, in the format http://[[user][:pass]@]host[:port]/, and then select Continue. 21 On the Configuring tasksel screen, select No automatic updates. 22 On the Software Selection screen, select OpenSSH server, and then select Continue. 23 On the Install the GRUB boot loader on a hard disk screeen, select Yes to install the GRUB boot loader. 24 When Finish the Installation screen appears, remove the Ubuntu image CD-ROM or USB memory stick, so that the computer boots from the newly installed operating system. 25 On the Finish the Installation screen, select Continue to finish the installation.

Extreme Networks OpenStack Plugin 2.0 Installation Guide

11

4 Preparing the Servers and Switches

Updating the Servers Setting up the Ethernet Port for the Management Interface

Updating the Servers After installing Ubuntu, you need to update the servers: apt-get update

Setting up the Ethernet Port for the Management Interface On every server, each physical Ethernet port must be identified (which one is eth0, eth1, eth2, etc.). This is important since the setup script (see Creating Tenants on page 35) uses particular Ethernet interfaces for data, control, storage, and management. Note If you do not know the port order on the server, plug a cable into each port and run either dmesg or ethtool to identify each port. If you want to change the port order, edit the following file: /etc/udev/rules.d/70-persistent-net.rules. Eth0 is used for the management interface. To set up Eth0 with an IP address and DNS: 1

Set up an IP address for Eth0:

ifconfig eth0 x.x.x.x/x 2 Set up DNS by adding the following to Eth0 inet static in /etc/network/interfaces: dns-nameservers x.x.x.x 3 Define subnets for the management and control networks.

5 Installing OpenStack Plugin 2.0 on the Servers

Downloading and Preparing OpenStack Installing OpenStack on the OSController Installing OpenStack on the OSHosts

Downloading and Preparing OpenStack To download and prepare OpenStack 2.0 on all servers: 1

Download the OpenStack 2.0 software package: Note After purchasing a license from Extreme Networks, you are provided with a URL to download the software.

2 Copy the package onto the host under the stack user's home directory (/home/stack). 3 Log in as the stack user and untar the package: stack@OSController$ tar xvf extr_openstack_v200bXX.tar.gz Where XX is the software build number.

Installing OpenStack on the OSController Note The OScontroller must be up and running prior to stacking any compute nodes. To install OpenStack 2.0 on OSController: 1

On the host, edit the setup.config file—located in /home/stack/ extr_openstack_v200bXX/setup ( where XX is the software build number), assuming the tar file was extracted in the users (stack) home directory . The setup script prepares the OSController. The management, control, data, and storage interfaces are defined (two interfaces are bonded together). Thus, BondH0=Ctrl Bond1=Data Bond2=Storage. If a single interface is defined, no bond is created. If you make a mistake, you can edit the file again, and then re-run the setup script. stack@OSController$ cd extr_openstack_v200bXX/setup stack@OSController$ vi setup.config Where XX is the software build number.

Installing OpenStack Plugin 2.0 on the Servers

The following is a sample of the setup.config file on the OSController node. All parts that need to be edited are in bold. ########################################################################## # # config file for extreme_setup_controller.sh and extreme_setup_compute.sh # ########################################################################## ############################################ # env variables specific to this host ############################################ # Network interfaces: # - Management network. This is an out-of-band network used for install # and support. It should already be configured. It cannot be bonded. MGMT_IF=eth0 # - Control network. This is how Openstack talks to itself. It's also # where Openstack exposes its services. It will be configured by this # script. If two interfaces are provided, they will be bonded. CTRL_IF_1=eth1 CTRL_IF_2=eth2 # - Data network. This is where we'll layer tenant networks. It will be # configured by this script. If two interfaces are provided, they will # be bonded. DATA_IF_1=eth3 DATA_IF_2=eth4 # - Storage network. Simple network for storage traffic. It will be # configured by this script. STORAGE_IF_1=eth5 STORAGE_IF_2=eth6 # IP addresses: MGMT_IP=10.68.61.116 CTRL_IP=192.168.50.10

############################################ # env variables common for all hosts ############################################ # Networks: MGMT_NET=10.68.61.0 MGMT_MASK=255.255.255.0 CTRL_NET=192.168.50.0 CTRL_MASK=255.255.255.0 # Other networking DEFAULT_GATEWAY=10.68.61.1 DNS_NAMESERVERS=10.6.16.32 FLOATING_RANGE=192.168.50.0/27 #This will need to be changed to whatever your floating range will be - does not have to be overlapped with the ctrl net # Hostnames and control network IP addresses: CONTROLLER=OSController CONTROLLER_IP=192.168.50.10

Extreme Networks OpenStack Plugin 2.0 Installation Guide

14

Installing OpenStack Plugin 2.0 on the Servers

COMPUTE_HOST1=OSHost1 COMPUTE_HOST1_IP=192.168.50.9 COMPUTE_HOST2=OSHost2 COMPUTE_HOST2_IP=192.168.50.8 #COMPUTE_HOST3=OSHost3 #COMPUTE_HOST3_IP=192.168.9.13 #COMPUTE_HOST4=OSHost4 #COMPUTE_HOST4_IP=192.168.9.13 #COMPUTE_HOST5=OSHost5 #COMPUTE_HOST5_IP=192.168.9.13 # Override branches and repos. If you want the defaults, then leave it commented. # Use these if you are on the Corporate Network - Otherwise comment out all of this below UBUNTU_ROOT=http://10.68.61.14/ubuntu/201309131057/ PIP_MIRROR=http://10.68.61.14/pypi/201309120802/pypi/web GIT_BASE=http://buildmaster.extremenetworks.com/anon-gitrepo #DEVSTACK_BRANCH=extreme/grizzly DEVSTACK_URL=http://buildmaster.extremenetworks.com/anon-gitrepo/ openstack- dev/devstack.git #DEVSTACK_BRANCH=extreme/grizzly #DEVSTACK_URL=http://10.68.61.68/anon-git/devstack.git #OPENSTACK_BRANCH=2013.1.3 #CINDER_BRANCH=${OPENSTACK_BRANCH} #GLANCE_BRANCH=${OPENSTACK_BRANCH} #HORIZON_BRANCH=${OPENSTACK_BRANCH} #KEYSTONE_BRANCH=${OPENSTACK_BRANCH} #NOVA_BRANCH=${OPENSTACK_BRANCH} #QUANTUM_BRANCH=${OPENSTACK_BRANCH} #OPENSTACKCLIENT_BRANCH=0.2.1 #CINDERCLIENT_BRANCH=1.0.5 #GLANCECLIENT_BRANCH=0.6.0 #KEYSTONECLIENT_BRANCH=0.2.5 #NOVACLIENT_BRANCH=2.10.0 #QUANTUM_CLIENT_BRANCH=2.2.6

2 Run the setup script extreme_setup_controller.sh setup.config for the first time (may require a password). stack@OSController$ ./extreme_setup_controller.sh setup.config 3 If instructed by the setup script, reboot the machine.

Extreme Networks OpenStack Plugin 2.0 Installation Guide

15

Installing OpenStack Plugin 2.0 on the Servers

4 If you rebooted the machine in the previous step, re-run the setup script extreme_setup_controller.sh setup.config. stack@OSController$ cd extr_openstack_v200bXX/setup stack@OSController$ ./extreme_setup_controller.sh setup.config (upon successful installation) OSController has been successfully setup for bond0:192.168.50.230 Where XX is the software build number. This setup script automates the following steps required for an OpenStack controller host: • Sets up network interfaces (with NIC bonding if applicable) for this server. • Downloads devstack and other required software packages. • Configures devstack/localrc for OpenStack components and OpenStack 2.0.

• •

Starts devstack and installs OpenStack packages. Configures NFS and libvirt for live migration.

Installing OpenStack on the OSHosts After the OSController is stacked and operational (see Installing OpenStack on the OSController on page 13), you can set up the compute nodes (OSHost1 and OSHost2). Note Installing OpenStack 2.0 deletes all of the initial network configuration on the compute nodes. To install the OpenStack 2.0 software package on OSHost1: 1

On this host, edit the setup.config file—located in /home/stack/ extr_openstack_v200bXX/setup ( where XX is the software build number), assuming the tar file was extracted in the users (stack) home directory . The setup script prepares OSHost1. The management, control, data, and storage interfaces are defined (two interfaces are bonded together). Thus, BondH0=Ctrl Bond1=Data Bond2=Storage. If a single interface is defined, no bond is created. If you make a mistake, you can edit the file again, and then re-run the setup script. stack@OSHost1$ cd /home/stack/extr_openstack_v200bXX/setup stack@OSHost1$ vi setup.config Where XX is the software build number. The following is a sample of the setup.config file on the OSHost1 node. All parts that need to be edited are in bold. ########################################################################## # # config file for extreme_setup_controller.sh and extreme_setup_compute.sh # ########################################################################## ############################################ # env variables specific to this host ############################################ # Network interfaces:

Extreme Networks OpenStack Plugin 2.0 Installation Guide

16

Installing OpenStack Plugin 2.0 on the Servers

# - Management network. This is an out-of-band network used for install # and support. It should already be configured. It cannot be bonded. MGMT_IF=eth0 # - Control network. This is how Openstack talks to itself. It's also # where Openstack exposes its services. It will be configured by this # script. If two interfaces are provided, they will be bonded. CTRL_IF_1=eth1 CTRL_IF_2=eth2 # - Data network. This is where we'll layer tenant networks. It will be # configured by this script. If two interfaces are provided, they will # be bonded. DATA_IF_1=eth3 DATA_IF_2=eth4 # - Storage network. Simple network for storage traffic. It will be # configured by this script. STORAGE_IF_1=eth5 STORAGE_IF_2=eth6 # IP addresses: MGMT_IP=10.68.61.116 CTRL_IP=192.168.50.10

############################################ # env variables common for all hosts ############################################ # Networks: MGMT_NET=10.68.61.0 MGMT_MASK=255.255.255.0 CTRL_NET=192.168.50.0 CTRL_MASK=255.255.255.0 # Other networking DEFAULT_GATEWAY=10.68.61.1 DNS_NAMESERVERS=10.6.16.32 FLOATING_RANGE=192.168.50.0/27 #This will need to be changed to whatever your floating range will be - does not have to be overlapped with the ctrl net # Hostnames and control network IP addresses: CONTROLLER=OSController CONTROLLER_IP=192.168.50.10 COMPUTE_HOST1=OSHost1 COMPUTE_HOST1_IP=192.168.50.9 COMPUTE_HOST2=OSHost2 COMPUTE_HOST2_IP=192.168.50.8 #COMPUTE_HOST3=OSHost3 #COMPUTE_HOST3_IP=192.168.9.13 #COMPUTE_HOST4=OSHost4 #COMPUTE_HOST4_IP=192.168.9.13 #COMPUTE_HOST5=OSHost5 #COMPUTE_HOST5_IP=192.168.9.13

Extreme Networks OpenStack Plugin 2.0 Installation Guide

17

Installing OpenStack Plugin 2.0 on the Servers

# Override branches and repos. If you want the defaults, then leave it commented. # Use these if you are on the Corporate Network - Otherwise comment out all of this below UBUNTU_ROOT=http://10.68.61.14/ubuntu/201309131057/ PIP_MIRROR=http://10.68.61.14/pypi/201309120802/pypi/web GIT_BASE=http://buildmaster.extremenetworks.com/anon-gitrepo #DEVSTACK_BRANCH=extreme/grizzly DEVSTACK_URL=http://buildmaster.extremenetworks.com/anon-gitrepo/ openstack- dev/devstack.git #DEVSTACK_BRANCH=extreme/grizzly #DEVSTACK_URL=http://10.68.61.68/anon-git/devstack.git #OPENSTACK_BRANCH=2013.1.3 #CINDER_BRANCH=${OPENSTACK_BRANCH} #GLANCE_BRANCH=${OPENSTACK_BRANCH} #HORIZON_BRANCH=${OPENSTACK_BRANCH} #KEYSTONE_BRANCH=${OPENSTACK_BRANCH} #NOVA_BRANCH=${OPENSTACK_BRANCH} #QUANTUM_BRANCH=${OPENSTACK_BRANCH} #OPENSTACKCLIENT_BRANCH=0.2.1 #CINDERCLIENT_BRANCH=1.0.5 #GLANCECLIENT_BRANCH=0.6.0 #KEYSTONECLIENT_BRANCH=0.2.5 #NOVACLIENT_BRANCH=2.10.0 #QUANTUM_CLIENT_BRANCH=2.2.6

2 Run the setup script extreme_setup_compute.sh setup.config for the first time (may require a password). stack@OSHost1$ ./extreme_setup_compute.sh setup.config. 3 If instructed by the setup script, reboot the machine. 4 If you rebooted the machine in the previous step, re-run the setup script extreme_setup_compute.sh setup.config. stack@OSHost1$ cd /home/stack/extr_openstack_v200bXX/setup/ stack@OSHost1$ ./extreme_setup_compute.sh setup.config (upon successful installation) OSHost1 has been successfully setup for bond0:192.168.50.231 Where XX is the software build number. 5 Repeat procedure for OSHost2. This setup script automates the following steps required for an OpenStack compute host:

• • • • •

Sets up network interfaces (with NIC bonding if applicable) for this server. Downloads devstack and other required software packages. Configures devstack/localrc for OpenStack components and OpenStack 2.0. Starts devstack and installs OpenStack packages. Configures NFS and libvirt for live migration.

Extreme Networks OpenStack Plugin 2.0 Installation Guide

18

6 Configuring Extreme Networks Switches

Configuring Control and TOR Switches Configuring the TOR1 Switch Configuring the TOR2 Switch Configuring the Control 1 Switch Configuring the Control 2 Switch

Configuring Control and TOR Switches For all the servers to communicate with each other, the TOR and control switches need to be setup correctly. The TOR switches also need certain configurations applied to work correctly with OpenStack 2.0. The easiest way to accomplish this is use a default.xsf file on each switch, and then run the command unconfig switch all. This way the switches are configured correctly each time. There are also loops in the control topology. This is prevented by using a UPM script, a software redundant port, or simply disabling one port on each TOR switch to ensure a loop-free control network.

Configuring the TOR1 Switch Since OpenStack 2.0. uses virtual routers, all ports are deleted from virtual router vr-default, except for the ctrl-net ports. You can create a separate ctrl-net virtual router if needed. Only port 1 is added to the ctrl-net VLAN to prevent a loop. To configure the TOR1 switch: 1 Begin with the factory default configuration. 2 Edit default.xsf: vi default.xsf configure snmp sysname TOR1 configure default del ports all configure vlan mgmt ipaddress 10.68.61.226/24 configure iproute add default 10.68.61.1 vr vr-mgmt configure dns-client add name-server 10.6.16.32 vr vr-mgmt configure dns-client add name-server 10.6.17.21 vr vr-mgmt configure dns-client add name-server 10.6.25.30 vr vr-mgmt enable web http create vlan ctrl-net conf ctrl-net ipaddress 192.168.50.12/24 configure vlan ctrl-net add ports 1 untagged conf vr "VR-Default" del port 3 - 58 disable idletimeout

Configuring Extreme Networks Switches

Configuring the TOR2 Switch Since OpenStack 2.0. uses virtual routers, all ports are deleted from virtual router vr-default, except for the ctrl-net ports. You can create a separate ctrl-net vr if needed. Only port 1 is added to the ctrl-net VLAN to prevent a loop. To configure the TOR2 switch: 1 Begin with the factory default configuration. 2 Edit default.xsf: vi default.xsf configure snmp sysname TOR2 configure default del ports all configure vlan mgmt ipaddress 10.68.61.227/24 configure iproute add default 10.68.61.1 vr vr-mgmt configure dns-client add name-server 10.6.16.32 configure dns-client add name-server 10.6.17.21 configure dns-client add name-server 10.6.25.30 create vlan ctrl-net conf ctrl-net ipaddress 192.168.50.13/24 configure vlan ctrl-net add ports 1 untagged enable web http conf vr "VR-Default" del port 3 - 58 disable idletimeout enable cli-config-logging

Configuring the Control 1 Switch To configure the Control 1 switch: 1 Begin with the factory default configuration. 2 Edit default.xsf: vi default.xsf configure snmp sysname CTRL1 configure default del ports all configure vlan Mgmt ipaddress 10.68.61.224/24 configure iproute add default 10.68.61.1 vr VR-Mgmt create vlan ctrl-net conf "ctrl-net" tag 300 configure vlan ctrl-net ipaddress 192.168.50.10/24 enable sharing 19 grouping 19,20 algorithm address-based L2 conf "ctrl-net" add port 7-11,19 untagged

Configuring the Control 2 Switch To configure the Control 2 switch: 1

Begin with the factory default configuration.

Extreme Networks OpenStack Plugin 2.0 Installation Guide

20

Configuring Extreme Networks Switches

2 Edit default.xsf: vi default.xsf configure snmp sysname CTRL2 configure default del ports all configure vlan Mgmt ipaddress 10.68.61.225/24 configure iproute add default 10.68.61.1 vr VR-Mgmt create vlan ctrl-net conf "ctrl-net" tag 300 configure vlan ctrl-net ipaddress 192.168.50.11/24 enable sharing 19 grouping 19,20 algorithm address-based L2 conf "ctrl-net" add port 7-11,19 untagged

Extreme Networks OpenStack Plugin 2.0 Installation Guide

21

7 Starting and Stopping Extreme Networks OpenStack

Starting and Stopping Extreme Networks OpenStack Modifying the Devstack localrc Parameters Starting OpenStack on OSController Starting OpenStack on OSHost1 and OSHost2 Verifying the OSController and OSHosts Populating the Topology Database Configuring the Network Fabric—LAG/MLAG Shutting Down OpenStack on All Servers Logs

Starting and Stopping Extreme Networks OpenStack The Extreme Networks reference topology setup (see Reference Topology Setup on page 6) with OpenStack 2.0 must be started up in a specific order: 1 Start up all of the TOR switches. 2 (Optional) Modify devstack's localrc parameters (see Modifying the Devstack localrc Parameters on page 23). 3 Start OpenStack on the controller (OSController) (see Starting OpenStack on OSController on page 24). 4 After the controller is completely up and running, start OpenStack on the compute hosts (OSHost1 and OSHost2) (see Starting OpenStack on OSHost1 and OSHost2 on page 25). 5 Populate the OpenStack 2.0 topology database from the controller (see Populating the Topology Database on page 26). 6 Configure the network fabric (see Configuring the Network Fabric—LAG/MLAG on page 31). 7 Create sample tenants and tenant networks from the controller (see Creating Tenants on page 35). 8 Create/migrate/delete tenant VM instances (see Creating Tenant Virtual Machine Instances on page 44, Migrating Tenant Virtual Machine Instances (Live Migration) on page 50, and Deleting Tenant Virtual Machine Instances on page 51). 9 When finished, shut down OpenStack on all servers (see Shutting Down OpenStack on All Servers on page 33).

Starting and Stopping Extreme Networks OpenStack

Modifying the Devstack localrc Parameters Several configuration parameters for running OpenStack Nova/Neutron are configured from devstack's localrc (/home/stack/devstack/localrc). Extreme Networks has extended this localrc to include key parameters for OpenStack 2.0 along with the OVS plugin. Below is a list for parameters for OpenStack 2.0 that can be modified in localrc: Parameter

Default

Description

EXTR_VLAN_START * EXTR_VLAN_END

101 ** 4,000

Start ID of VLAN from tenant VMs Last ID of VLAN

EXTR_MAX_CORES

16

Maximum number of virtual machines (VMs) per compute host

To edit localrc and have the changes take effect: 1

Ensure that all servers are unstacked.

./unstack.sh 2 Install OSController: ./extreme_setup_controller.sh setup.config 3 Change the parameters in localrc: vi /opt/stack/devstack/localrc 4 Unstack OSController: stack@OSController:/opt/stack/devstack$ ./unstack.sh 5 Re-stack OSController: stack@OSController:/opt/stack/devstack$ ./stack.sh 6 Install OSHost1 and OSHost2: ./extreme_setup_compute.sh setup.config Advanced Information: If OpenStack 2.0 is to be plugged into OpenStack manually (by other means than using devstack), you must provide the above configuration parameters in a quantum .ini file: /etc/quantum/plugins/ extreme/extreme_quantum_plugin.ini. [VIRT_NETWORK] # Virtual network plugin responsible for virtual network connectivity. vplugin = quantum.plugins.openvswitch.ovs_quantum_plugin.OVSQuantumPluginV2 [SCHEDULER] # Scheduler for VM instances. scheduler = quantum.plugins.extreme.scheduler.topology_aware_scheduler.TopologyAwareS cheduler scheduler_vman = * **

VLAN/CVID/SVID of 1 is reserved on ExtremeXOS. Default values for VLAN/CVID/SVID used in this guide are only intended for ease of demonstration and have no particular meanings.

Extreme Networks OpenStack Plugin 2.0 Installation Guide

23

Starting and Stopping Extreme Networks OpenStack

quantum.plugins.extreme.scheduler.topology_aware_scheduler_vman.TopologyA wareSchedulerVman [GLOBAL_PREFS] multi_tenants_per_host=True multi_tenancy_solution_type=VLAN min_vlan=101 max_vlan=4000 persistant_cfg=True Additionally, OpenStack 2.0 must be specified as a core plugin in quantum conf: /etc/quantum/ quantum.conf. # Quantum plugin provider module core_plugin = quantum.plugins.extreme.api.extreme_quantum_plugin_v2.ExtremeQuantumPlugi nV2 Also, OpenStack 2.0's Nova components need to be specified in nova conf: /etc/nova/nova.conf. compute_scheduler_driver=quantum.plugins.extreme.nova.extreme_scheduler_v 2.ExtrSchedulerV2 network_api_class=quantum.plugins.extreme.nova.quantumv2_api.API For details about OpenStack configuration parameters, see the following files, which are controlled from devstack's localrc and cannot be modified directly:

• • • •

Nova:/etc/nova/nova.conf Quantum:/etc/quantum/quantum.conf OVS Quantum Plugin:/etc/quantum/plugins/openvswitch/ovs_quantum_plugin.ini OpenStack 2.0: /etc/quantum/plugins/extreme/extreme_qunatum_plugin.ini

Starting OpenStack on OSController Note Several configuration parameters for running OpenStack Nova/Neutron are configured from devstack's localrc (/home/stack/devstack/localrc). Extreme Networks has extended this localrc to include key parameters for OpenStack 2.0 along with the OVS plugin. If you need to change these parameters, do so before starting OpenStack on the OSController (see Modifying the Devstack localrc Parameters on page 23). To start OpenStack on the OSController: Run stack.sh to start OpenStack controller with the above configurations in localrc: stack@OSController$ cd /opt/stack/devstack stack@OSController$ ./stack.sh (upon successful start) Horizon is now available at http://192.168.9.11/ Keystone is serving at http://192.168.9.11:5000/v2.0/ Examples on using novaclient command line is in exercise.sh The default users are: admin and demo The password: nova123 This is your host ip: 192.168.9.11 stack.sh completed in 91 seconds.

Extreme Networks OpenStack Plugin 2.0 Installation Guide

24

Starting and Stopping Extreme Networks OpenStack

Starting OpenStack on OSHost1 and OSHost2 To start OpenStack on OSHost1 and OSHost2: After successfully starting the OSController, run stack.sh to start OpenStack computes (you do not need to modify localrc on compute hosts). Note Always start the controller first before starting any compute hosts. stack@OSHost1$ cd /opt/stack/devstack stack@OSHost1$ ./stack.sh (upon successful launch) This is your host ip: 192.168.9.12 stack.sh completed in 19 seconds. ----------------------------------------------stack@OSHost2$ cd /opt/stack/devstack stack@OSHost2$ ./stack.sh (upon successful launch) This is your host ip: 192.168.9.13 stack.sh completed in 19 seconds.

Verifying the OSController and OSHosts After starting OpenStack on OSController and OSHosts, you can verify that they are setup properly. To verify OSController and OSHosts setup: 1

To grant permission to execute commands using OpenStack Compute (Nova), Networking (Neutron), and Identity Service (Keystone): source /opt/stack/devstack/openrc admin admin Note If interested, read the openrc file.

Extreme Networks OpenStack Plugin 2.0 Installation Guide

25

Starting and Stopping Extreme Networks OpenStack

2 Type the following commands to view the status of OSController and OSHost1/OSHost2: stack@OSController:~/extr_openstack_v200b20/setup$ nova service-list +-----------------+-------------+---------+--------+-----+--------------------------+ | Binary | Host | Zone | Status | State| Updated_at | +-----------------+-------------+---------+--------+-----+--------------------------+ | nova-conductor | OSController| internal| enabled| up | 2014-01-03T20:25:48.000000| | nova-cert | OSController| internal| enabled| up | 2014-01-03T20:25:50.000000| | nova-scheduler | OSController| internal| enabled| up | 2014-01-03T20:25:52.000000| | nova-consoleauth| OSController| internal| enabled| up | 2014-01-03T20:25:46.000000| | nova-compute | OSHost1 | nova | enabled| up | 2014-01-03T20:25:49.000000| | nova-compute | OSHost2 | nova | enabled| up | 2014-01-03T20:25:54.000000| +-----------------+-------------+---------+--------+-----+--------------------------+ stack@OSController:~/extr_openstack_v200b20/setup$ nova host-list +--------------+-------------+----------+ | host_name | service | zone | +--------------+-------------+----------+ | OSController | conductor | internal | | OSController | cert | internal | | OSController | scheduler | internal | | OSController | consoleauth | internal | | OSHost1 | compute | nova | | OSHost2 | compute | nova | +--------------+-------------+----------+

Populating the Topology Database To use OpenStack 2.0, the topology database must be populated first with the proper topology information—the actual physical topology. The key topology information includes:

• • • • • •

Zones and pods information. [zone / pod] Hostnames of all the servers (OSController, Network and Compute) and their roles. [server_host] L2 devices (TOR switches) login information. [device] L2 device slot/port information. [dev_port] Inter-connection between servers and TOR switches [server_device_connect] Inter-connection between TORs and other devices. [dev_interconnect]

You can create a new script or edit an existing sample script. Sample scripts are available under the / home/stack/extr_openstack_v200bXX/setup/configs directory (where XX is the build

Extreme Networks OpenStack Plugin 2.0 Installation Guide

26

Starting and Stopping Extreme Networks OpenStack

number). Choose an example that is similar to your topology and modify it. The following procedure shows how to edit an existing script. Note You must run the script on OSController every time devstack (stack.sh) is started, and before creating virtual networks. To edit an existing sample json script to update the topology database: 1

On OSController, navigate to the folder with the sample json scripts: cd /home/stack/extr_openstack_v200bXX/setup/configs

Where XX is the build number. 2 Copy and edit a sample json script (in this example, bens_one_tier_two_tor_topology.json): stack@OSController:~/extr_openstack_v200bXX/setup/configs$ ls create_tenant.json create_tenant_vr.json one_tier_one_tor.json one_tier_two_tor.json openstack_init.json cp one_tier_two_tor.json bens_one_tier_two_tor_topology.json vi bens_one_tier_two_tor_topology.json 3 Modify the hosts part of the json file to define the servers and appliances in your OpenStack pod. Ensure that the name matches the name on the server/appliance and that the number of ports/NICs is defined. (These are the data NICs that make up the bond on the servers, and the single NIC for border gateway devices.) "hosts": [ { "name":"OSController", "type":"network", "num_of_nics":"2" }, { "name":"OSHost1", "type":"compute", "num_of_nics":"2" }, { "name":"OSHost2", "type":"compute", "num_of_nics":"2" }, { "name":"ose-bgw1", "type":"appliance", "num_of_nics":"1" }, { "name":"ose-bgw2", "type":"appliance", "num_of_nics":"1" } ], Note Ose-bgw1 and osebgw2 are not Extreme Networks devices, but instead "appliances" (routers, firewalls, etc.) that plug into the TOR switches and then forward the VM traffic.

Extreme Networks OpenStack Plugin 2.0 Installation Guide

27

Starting and Stopping Extreme Networks OpenStack

4 Modify each TOR switch's information to include the appropriate management (control network) IP address and the ports used for the connections between the TOR switches, the servers, and the appliances: Note In this example: • TOR1 has an IP address of 192.168.50.12, has ports 7,3,5,8,4,6 going to the servers, port 46 going to the appliance (Border Gateway), and ports 19 and 20 configured in a LAG between the TOR switches. • TOR2 has an IP address of 192.168.50.13, has ports 7,3,5,8,4,6 going to the servers, port 46 going to the appliance (Border Gateway), and ports 19 and 20 configured in a LAG between the TOR switches. If the port is used for the storage network, include “storage_port”:”1”; if the port is a nonmaster part of the LAG, add “master_port”:”0” "devices": [ { "name":"TOR1", "mgmt_ip_addr":"192.168.50.12", "default_vlan":"Default", "slots" : [ { "ports": [ {"port_id":"7"}, {"port_id":"3"}, {"port_id":"5"}, {"port_id":"19", "storage_port":"1"}, {"port_id":"20", "master_port":"0","storage_port":"1"}, {"port_id":"8", "storage_port":"1"}, {"port_id":"4", "storage_port":"1"}, {"port_id":"6", "storage_port":"1"}, {"port_id":"46"} ] } ] }, { "name":"TOR2", "mgmt_ip_addr":"192.168.50.13", "default_vlan":"Default", "slots" : [ { "ports": [ {"port_id":"8"}, {"port_id":"4"}, {"port_id":"6"}, {"port_id":"19", "storage_port":"1"}, {"port_id":"20", "master_port":"0","storage_port":"1"}, {"port_id":"7", "storage_port":"1"}, {"port_id":"3", "storage_port":"1"}, {"port_id":"5", "storage_port":"1"}, {"port_id":"46"} ] } ] } ],

Extreme Networks OpenStack Plugin 2.0 Installation Guide

28

Starting and Stopping Extreme Networks OpenStack

5 Specify the authentication to the switches. In this example, the username is admin, and the password is blank. device_auth_info": {"user_name":"admin", "password":""},

Extreme Networks OpenStack Plugin 2.0 Installation Guide

29

Starting and Stopping Extreme Networks OpenStack

6 Finally, identify the connections between the devices with the device name, followed by the port number. Notice how the LAG has two physical connections in this example: • Port 7 on TOR1 is connected to one of the two OSController data network ports • Port 8 on TOR2 is connected to the other OSController data network port • Port 3 on TOR1 is connected to one of the two OSHost1 data network ports • Port 4 on TOR2 is connected to the other OSHost1 data network port • Port 5 on TOR1 is connected to one of the two OSHost2 data network ports • Port 6 on TOR2 is connected to the other OSHost2 data network port • Port 19 on TOR1 is connected as a master to port 19 on TOR2 • Port 20 on TOR1 is connected as a slave to port 20 on TOR2 • Port 46 on TOR1 is connected to Border Gateway appliance 1 • Port 46 on TOR2 is connected to Border Gateway appliance 2 "connections": [ {"left":"device:TOR1:7","right":"host:OSController"}, {"left":"device:TOR2:8","right":"host:OSController"}, {"left":"device:TOR1:3","right":"host:OSHost1"}, {"left":"device:TOR2:4","right":"host:OSHost1"}, {"left":"device:TOR1:5","right":"host:OSHost2"}, {"left":"device:TOR2:6","right":"host:OSHost2"}, {"left":"device:TOR1:19","right":"device:TOR2:19"}, {"left":"device:TOR1:20","right":"device:TOR2:20"}, {"left":"device:TOR1:48","right":"host:ose-bgw1"}, {"left":"device:TOR2:48","right":"host:ose-bgw2"} ] After you finish editing the file, it should resemble the following. The following json script is based on the reference topology (see Reference Topology Setup on page 6): { "zones": [ { "name": "zone1", "pods": [ { "name": "pod1", "hosts": [ { "name": "OSController", "type": "network", "num_of_nics": "2" }, { "name": "OSHost1", "type": "compute", "num_of_nics": "2" }, { "name": "OSHost2", "type": "compute", "num_of_nics": "2" }, { "name": "ose-bgw1", "type": "appliance", "num_of_nics": "1" }, { "name": "ose-bgw2", "type": "appliance", "num_of_nics": "1" } Extreme Networks OpenStack Plugin 2.0 Installation Guide ],

30

Starting and Stopping Extreme Networks OpenStack

7 Now that the information represents the physical topology, the database can be populated. Run the following: cd .. stack@OSController:~/extr_openstack_v200b17/setup/$ pwd /home/stack/extr_openstack_v200b17/setup/ stack@OSController:~/extr_openstack_v200b17/setup/$ source /opt/stack/ devstack/openrc admin admin stack@OSController:~/extr_openstack_v200b17/setup/topology$./ extreme_prep_topology.py -i configs/ bens_one_tier_two_tor_topology.json Topology file: configs/bens_one_tier_two_tor_topology.json Loading topology description...done. Cleaning current topology database...done Creating topology... Creating zone:zone1 Creating pod:pod1 Creating host:OSController Creating host:OSHost1 Creating host:OSHost2 Creating host:ose-bgw1 Creating host:ose-bgw2 Creating device:TOR1 nslots: 1 nports: 9 Creating device:TOR2 nslots: 1 nports: 9 Creating connection: device:TOR1:7host:OSController Creating connection: device:TOR2:8host:OSController Creating connection: device:TOR1:3host:OSHost1 Creating connection: device:TOR2:4host:OSHost1 Creating connection: device:TOR1:5host:OSHost2 Creating connection: device:TOR2:6host:OSHost2 Creating connection: device:TOR1:19device:TOR2:19 Creating connection: device:TOR1:20device:TOR2:20 Creating connection: device:TOR1:46host:ose-bgw1 Creating connection: device:TOR2:46host:ose-bgw2

Configuring the Network Fabric—LAG/MLAG After the topology database is prepared, the next step is to run extreme_prep_openstack.py using config/openstack_init.json as an input. The openstack_init.json script sets up the initial fabric on the TOR switches. The storage VLAN is created and sharing is enabled on the ISC links, plus server ports. Additionally, the script configures MLAG on both the TOR1 and TOR2 switches, complete with the ISC, and enables MLAG on the ports facing the OSController and compute nodes. A dynamic ACL is also created to block VRRP advertisements to make VRRP active/active on the TOR1 and TOR2 switches. To configure the network fabric on the TOR switches:

Extreme Networks OpenStack Plugin 2.0 Installation Guide

31

Starting and Stopping Extreme Networks OpenStack

1

Change the subnet, if needed, in the openstack_init.json file. The openstack_init.json file describes the external network that virtual machines can use. This is a VLAN called PUBLIC on the switch:

stack@OSController:~/extr_openstack_v200b17/setup$ cat configs/ openstack_init.json // // This file contains a json description for an extrnal // network to be created by the extreme_prep_openstack.py // script. This example creates a public network to which // tenants within the openstack environment can connect // for external access. // { "extnets" : [ { "name": "PUBLIC", "owner": "admin", "subnets" : [ { "name": "PUBLIC-subnet", "gateway": "192.168.24.1", "range": "192.168.24.0/24" } ] } ] } 2 Run the script: stack@OSController:~/extr_openstack_v200bXX/setup$ ./ extreme_prep_openstack.py -i configs/openstack_init.json OpenStack init file: configs/openstack_init.json Loading initial OpenStack objects description...done. Creating initial OpenStack objects... Creating extnet:PUBLIC Creating subnet:PUBLIC-subnet 192.168.24.1 192.168.24.0/24 Where XX is the build number. The following excerpts from the script extreme_prep_openstack.py show the TOR switch configuration changes: TOR1 switch configuration additions: # # Module vlan configuration. # create vlan "isc-vlan" configure vlan isc-vlan tag 4093 create vlan "storage-vlan" configure vlan storage-vlan tag 4094 enable sharing 19 grouping 19-20 algorithm address-based L2 lacp enable sharing 7 grouping 7 algorithm address-based L2 lacp enable sharing 3 grouping 3 algorithm address-based L2 lacp enable sharing 5 grouping 5 algorithm address-based L2 lacp configure vlan isc-vlan add ports 19 tagged configure vlan storage-vlan add ports 4, 6, 8, 19 tagged configure vlan isc-vlan ipaddress 1.1.1.1 255.255.255.0 # # Module acl configuration.

Extreme Networks OpenStack Plugin 2.0 Installation Guide

32

Starting and Stopping Extreme Networks OpenStack

# create access-list NO-ISC-VRRP-TOR1 " destination-address 224.0.0.18/32 ;" " deny ;" application "Cli" configure access-list add NO-ISC-VRRP-TOR1 last priority 0 zone SYSTEM ports 19 ingress TOR2 switch configuration additions: # # Module vlan configuration. # create vlan "isc-vlan" configure vlan isc-vlan tag 4093 create vlan "storage-vlan" configure vlan storage-vlan tag 4094 enable sharing 19 grouping 19-20 algorithm address-based L2 lacp enable sharing 8 grouping 8 algorithm address-based L2 lacp enable sharing 4 grouping 4 algorithm address-based L2 lacp enable sharing 6 grouping 6 algorithm address-based L2 lacp configure vlan isc-vlan add ports 19 tagged configure vlan storage-vlan add ports 3, 5, 7, 19 tagged configure vlan isc-vlan ipaddress 1.1.1.2 255.255.255.0 # # Module acl configuration. # create access-list NO-ISC-VRRP-TOR2 " destination-address 224.0.0.18/32 ;" " deny ;" application "Cli" configure access-list add NO-ISC-VRRP-TOR2 last priority 0 zone SYSTEM ports 19 ingress configure access-list add NO-ISC-VRRP-TOR2 last priority 0 zone SYSTEM ports 20 ingress # # Module vsm configuration. # configure mlag ports convergence-control fast create mlag peer "TOR1" configure mlag peer "TOR1" ipaddress 1.1.1.1 vr VR-Default enable mlag port 4 peer "TOR1" id 2 enable mlag port 6 peer "TOR1" id 3 enable mlag port 8 peer "TOR1" id 1

Shutting Down OpenStack on All Servers You should shut down OpenStack after you have finished using it or before restarting it (for example, after a configuration change). Run unstack.sh on all servers: stack@OSController$ cd /opt/stack/devstack stack@OSController$ ./unstack.sh

Logs Devstack generates log files for each OpenStack components in /opt/stack/logs/. You can find useful log information:

Extreme Networks OpenStack Plugin 2.0 Installation Guide

33

Starting and Stopping Extreme Networks OpenStack

• • • • • • • •

Devstack: /opt/stack/logs/stack.sh.log Nova API: /opt/stack/logs/screen/screen-n-api.log Nova Compute: /opt/stack/logs/screen/screen-n-cpu.log Nova Scheduler: /opt/stack/logs/screen/screen-n-sch.log Quantum Server: /opt/stack/logs/screen/screen-q-svc.log Quantum Agents: /opt/stack/logs/screen/screen-q-agt.log Quantum Agents: /opt/stack/logs/screen/screen-q-dhcp.log Quantum Agents: /opt/stack/logs/screen/screen-q-l3.log

Extreme Networks OpenStack Plugin 2.0 Installation Guide

34

8 Managing Tenants and Virtual Machines

Creating Tenants Creating Tenants Using Python Script and Configuration File (L3 Agent) Creating Tenants Using Python Script and Configuration File (Virtual Routers) Verifying TOR Switch Configuration after Tenant Creation (L3 Agent) Verifying TOR Switch Configuration after Tenant Creation (Virtual Routers) Creating Tenant Virtual Machine Instances Migrating Tenant Virtual Machine Instances (Live Migration) Deleting Tenant Virtual Machine Instances

Creating Tenants After setting up the initial fabric (see Configuring the Network Fabric—LAG/MLAG on page 31), you can create tenants along with their respective networks. After creating tenants, you can then create virtual machines (see Creating Tenant Virtual Machine Instances on page 44). To create tenants, run the Python script extreme_create_tenant.py that takes as its input a configuration file. The script and configuration files are available in the Configs folder. There are two possible configuration files: • create_tenant.json—uses L3 agent-based router (see see Creating Tenants Using Python Script and Configuration File (L3 Agent) on page 35) create_tenant_vr.json—uses Extreme Networks' virtual router (see Creating Tenants Using • Python Script and Configuration File (Virtual Routers) on page 37)

Creating Tenants Using Python Script and Configuration File (L3 Agent) One option for creating tenants is with a configuration file using an L3 Agent with PUBLIC network used as the default gateway. To create tenants using a Python script with a configuration file (L3 agent) as the input:

Managing Tenants and Virtual Machines

1

Edit the create_tenant.jsonconfiguration file to reflect the tenants and networks that need to be created: stack@OSController: cd /home/stack/extr_openstack_v200bXX/setup/ configs stack@OSController: vi create_tenant.json Where XX is the build number. The following is the create_tenant.json configuration file. Edit it to match the configured topology.

// // This file provides a description of an initial // tenant configuration that can be ingested by the // the extreme_create_tenant.py script. // // Through this interface, tenants, their instances, // and associated internal and external networks can // be created without requiring a series of individual // OpenStack interface commands. // // This example uses an L3-agent based router. { "tenants": [ { "name": "tenant1", //Specify the tenant name "description": "tenant1 description", //Specify the tenant description // // List of VMs // "instances": [ //Begin VM definition { "name": "tenant1-vm1", //Specify VM name "image": "cirros-0.3.1-x86_64-uec", //Specify VM image "flavor": "m1.tiny", //Specify VM flavor // Networks connected to the VM "networks": [ "tenant1-int-net" //Specify the network for the VM. You will actually create a network later in this file ] }, { //Repeat for additional VMs "name": "tenant1-vm2", "image": "cirros-0.3.1-x86_64-uec", "flavor": "m1.tiny", // Networks connected to the VM "networks": [ "tenant1-int-net" ] } ], // // List of internal networks // "networks": [ //create the internal networks // Internal network on which the VMs will connect { "name": "tenant1-int-net", //specify the internal network name "subnets": [ { Extreme Networks OpenStack Plugin 2.0 Installation Guide 36 "gateway": "192.168.1.1", //specify the

Managing Tenants and Virtual Machines

2 Run the Python file extreme_create_tenant.py: stack@OSController:~/extr_openstack_v200bXX/setup$ ./ extreme_create_tenant.py -i configs/create_tenant.json Tenant file: configs/create_tenant.json Loading tenant description...done. Creating tenant tenant1 Creating t1user Creating internal network tenant1-int-net Creating subnet tenant1-int-subnet Creating router tenant1-router1 Creating server tenant1-vm1 Creating server tenant1-vm2 Where XX is the build number.

Creating Tenants Using Python Script and Configuration File (Virtual Routers) One option for creating tenants is with a configuration file using virtual routers (see the following figure).

Extreme Networks OpenStack Plugin 2.0 Installation Guide

37

Managing Tenants and Virtual Machines

Figure 6: Reference Topology with Border Gateways with Created Tenant (Virtual Routers) To create tenants using a Python script with a configuration file (virtual routers) as the input:

Extreme Networks OpenStack Plugin 2.0 Installation Guide

38

Managing Tenants and Virtual Machines

1

Edit the create_tenant_vr.jsonconfiguration file to reflect the tenants and networks that need to be created: stack@OSController: cd /home/stack/extr_openstack_v200bXX/setup/ configs stack@OSController: vi create_tenant_vr.json Where XX is the build number. The following is the create_tenant_vr.json configuration file. Edit it to match the configured topology. // // // // // // // // // // // {

This file provides a description of an initial tenant configuration that can be ingested by the the extreme_create_tenant.py script. Through this interface, tenants, their instances, and associated internal and external networks can be created without requiring a series of individual OpenStack interface commands. This example uses an Extreme VR based router.

"tenants": [ { "name": "tenant1", //Specify the tenant name "description": "tenant1 description", //Specify the tenant description // // List of VMs // "instances": [ //Begin VM definition { "name": "tenant1-vm1", //Specify VM name "image": "cirros-0.3.1-x86_64-uec", //Specify VM image "flavor": "m1.tiny", //Specify VM flavor // Networks connected to the VM "networks": [ "tenant1-int-net" //Specify the network for the VM. You will actually create a network later in this file ] }, { //Repeat for additional VMs "name": "tenant1-vm2", "image": "cirros-0.3.1-x86_64-uec", "flavor": "m1.tiny", // Networks connected to the VM "networks": [ "tenant1-int-net" ] } ], // // List of internal networks // "networks": [ //Create the networks // Internal network on which the VMs will connect { "name": "tenant1-int-net", //Specify the network name "extr:vrrp": true, //"extr:vrrp": true flag makes the network internal Extreme Networks OpenStack Plugin 2.0 Installation Guide 39 "subnets": [

Managing Tenants and Virtual Machines

2 Run the Python file extreme_create_tenant.py: extreme_create_tenant.py -i configs/create_tenant_vr.json

Verifying TOR Switch Configuration after Tenant Creation (L3 Agent) After creating a tenant using an L3 Agent with PUBLIC network used as the default gateway (see Creating Tenants Using Python Script and Configuration File (L3 Agent) on page 35), you should make sure that the TOR switches were configured correctly. After using the L3 agent method, there should be a single VLAN across to the border gateways, and the PUBLIC VLAN should be connected to the L3 agent.

Extreme Networks OpenStack Plugin 2.0 Installation Guide

40

Managing Tenants and Virtual Machines

Show VLAN on TOR switches: * OS-Tor3.9 # sh vlan -------------------------------------------------------------------------------------------Name VID Protocol Addr Flags Proto Ports Virtual Active router / Total -------------------------------------------------------------------------------------------Default 1 -----------------------------------------------ANY 0 /0 VR-Default isc-vlan 4093 1.1.1.2 /24 ------I--------------------ANY 1 /1 VR-Default Mgmt 4095 10.68.67.23 /24 ---------------------------ANY 1 /1 VR-Mgmt OSMgmt 4092 192.168.9.23 /24 ---------------------------ANY 1 /2 VR-OSMgmt PUBLIC 4091 -----------------------------------------------ANY 2 /2 VR-Default storage-vlan 4094 -----------------------------------------------ANY 4 /4 VR-Default tenant1-int-net-101 101 ------------------------------------------------ ANY 3 /3 VRDefault -------------------------------------------------------------------------------------------Flags : (B) BFD Enabled, (c) 802.1ad customer VLAN, (C) EAPS Control VLAN, (d) Dynamically created VLAN, (D) VLAN Admin Disabled, (e) CES Configured, (E) ESRP Enabled, (f) IP Forwarding Enabled, (F) Learning Disabled, (i) ISIS Enabled, (I) Inter-Switch Connection VLAN for MLAG, (k) PTP Configured, (l) MPLS Enabled, (L) Loopback Enabled, (m) IPmc Forwarding Enabled, (M) Translation Member VLAN or Subscriber VLAN, (n) IP Multinetting Enabled, (N) Network Login VLAN, (o) OSPF Enabled, (O) Flooding Disabled, (p) PIM Enabled, (P) EAPS protected VLAN, (r) RIP Enabled, (R) Sub-VLAN IP Range Configured, (s) Sub-VLAN, (S) Super-VLAN, (t) Translation VLAN or Network VLAN, (T) Member of STP Domain, (v) VRRP Enabled, (V) VPLS Enabled, (W) VPWS Enabled, (Z) OpenFlow Enabled Total number of VLAN(s) : 7

Extreme Networks OpenStack Plugin 2.0 Installation Guide

41

Managing Tenants and Virtual Machines

Verifying TOR Switch Configuration after Tenant Creation (Virtual Routers) After creating a tenant using virtual routers (see Creating Tenants Using Python Script and Configuration File (Virtual Routers) on page 37), you should make sure that the TOR switches were configured correctly. After using the virtual router method, there should be three VLANs created on TOR1 and three on TOR2 in support of a single tenant network. As seen below on TOR1, VLANs T1-bgw1net-102, T1-bgw3net-104, and T1-net-101 were created within the VR-T1-router virtual router. T1-net-101 is the VLAN that extends to the servers—it has the port of the OSController, but no other server (yet). When a virtual machine (VM) is created, the server that houses that particular VM will then have its port added on the switch. The T1-bgw1net-102 VLAN extends to the Border Gateways and will have port 46 added in this case (see Reference Topology Setup on page 6). T1-bgw3net-104 is used as a redundancy VLAN—if the Border Gateway on TOR1 goes down, there is a route within the virtual router to route the traffic through this VLAN to the TOR2 switch and use its Border Gateway.

Extreme Networks OpenStack Plugin 2.0 Installation Guide

42

Managing Tenants and Virtual Machines

1

Show VLANs on TOR switches: * TOR1.2 # sh vlan ---------------------------------------------------------------------Name VID Protocol Addr Flags Proto Ports Virtual router Active/Total ---------------------------------------------------------------------ctrl-net 4092 192.168.50.12/24---- ANY 2 /2 VR-Default Default 1 -------------------- ANY 0 /0 VR-Default isc-vlan 4093 1.1.1.1 /24 I ANY 1 /1 VR-Default Mgmt 4095 10.68.61.226 /24---- ANY 1 /1 VR-Mgmt PUBLIC 4091 -------------------- ANY 3 /3 VR-Default storage-vlan 4094 -------------------- ANY 3 /4 VR-Default T1-bgw1net-102 102 30.0.0.1 /24 -f--- ANY 1 /1 VR-T1-router T1-bgw3net-104 104 50.0.0.1 /24 -f--- ANY 1 /1 VR-T1-router T1-net-101 101 10.0.0.3 /24 -f v- ANY 2 /2 VR-T1-router ---------------------------------------------------------------------* TOR2.2 # sh vlan ----------------------------------------------------------------------Name VID Protocol Addr Flags Proto Ports Virtual Router Active/Total ----------------------------------------------------------------------ctrl-net 4092 192.168.50.13 /24 ---- ANY 1 /1 VRDefault Default 1 ---------------------- ANY 0 /0 VRDefault isc-vlan 4093 1.1.1.2 /24 I ANY 1 /1 VRDefault Mgmt 4095 10.68.61.227 /24 ----- ANY 1 /1 VRMgmt PUBLIC 4091 ----------------------- ANY 3 /3 VRDefault storage-vlan 4094 ----------------------- ANY 4 /4 VRDefault T1-bgw2net-103 103 40.0.0.1 /24 -f ANY 1 /1 VR-T1router T1-bgw3net-104 104 50.0.0.2 /24 -f ANY 1 /1 VR-T1router T1-net-101 101 10.0.0.4 /24 -f--v ANY 2 /2 VR-T1router

Extreme Networks OpenStack Plugin 2.0 Installation Guide

43

Managing Tenants and Virtual Machines

2 Display the contents of the IP routing tables: * TOR1.7 # sh iproute vr "VR-T1-router" Ori Destination Gateway Mtr Flags #s Default Route 30.0.0.10 1 UG---S-um--f0h:0m:24s s Default Route 50.0.0.2 2 UG---S-um---0h:0m:23s #d 10.0.0.0/24 10.0.0.3 1 U------um--f0m:40s #d 30.0.0.0/24 30.0.0.1 1 U------um--f0h:0m:33s #d 50.0.0.0/24 50.0.0.1 1 U------um--f0h:0m:28s * TOR2.1 # sh iproute vr "VR-T1-router" Ori Destination Gateway Mtr Flags #s Default Route 40.0.0.10 1 UG---S-um--f0h:0m:46s s Default Route 50.0.0.1 2 UG---S-um---0h:0m:45s #d 10.0.0.0/24 10.0.0.4 1 U------um--f1m:1s #d 40.0.0.0/24 40.0.0.1 1 U------um--f0h:0m:54s #d 50.0.0.0/24 50.0.0.2 1 U------um--f0h:0m:50s

VLAN Duration T1-bgw1net-102 0d: T1-bgw3net-104 0d: T1-net-101 0d:0h: T1-bgw1net-102 0d: T1-bgw3net-104 0d:

VLAN Duration T1-bgw2net-103 0d: T1-bgw3net-104 0d: T1-net-101 0d:0h: T1-bgw2net-103 0d: T1-bgw3net-104 0d:

Creating Tenant Virtual Machine Instances The OpenStack is now ready to manage the lifecycle of tenant virtual machine (VM) instances. As a tenant (or admin), it allows creating new and deleting (see Deleting Tenant Virtual Machine Instances on page 51) existing VM instances. As an admin, it also allows migrating an active VM instance to a new server (see Migrating Tenant Virtual Machine Instances (Live Migration) on page 50).

Extreme Networks OpenStack Plugin 2.0 Installation Guide

44

Managing Tenants and Virtual Machines

1

In a web browser on any machine, connect to the Horizon Dashboard at 192.168.9.11. The Log In window appears. Horizon is a web-based GUI for OpenStack. It runs on the OpenStack OSController and connects to it via the control-net IP address of the OSController.

Figure 7: Horizon Login Window

Extreme Networks OpenStack Plugin 2.0 Installation Guide

45

Managing Tenants and Virtual Machines

2 Type login credentials in the User Name and Password boxes. Log in as:

• •

Tenant—When the Python script (see Creating Tenants on page 35) creates a tenant it provides a user name and password that can be used for that tenant. Logging in as a tenant provides access only to that tenant's resources. Admin—Logging in as admin allows you to access all tenants. User Name = admin; Password = nova123.

The Instances window appears.

Figure 8: Instances Window 3 On the left panel, click the Project tab. 4 Click Instances.

Extreme Networks OpenStack Plugin 2.0 Installation Guide

46

Managing Tenants and Virtual Machines

5 Click the Launch Instance button. The Launch Instance window appears.

Figure 9: Launch Instance Window 6 From the Image drop-down list, select an image. 7 In the Instance Name box, type a name. 8 Click the Networking tab. The Launch Instance—Networking Tab window appears.

Figure 10: Launch Instance—Networking Tab Window

Extreme Networks OpenStack Plugin 2.0 Installation Guide

47

Managing Tenants and Virtual Machines

9 Select T1-net. This is the tenants VLAN within the T1 virtual router. The other VLANs connect to the Border Gateways. Figure 11: Launch Instances Window with T1-net Selected

T1-net now appears under the heading Selected Networks. 10 Click the Launch button The Instances window appears showing the progress of creating the instance "Test1".

Figure 12: Instances Window with Instance "Test1" Creation in Progress After the instance is created, it should have its name and an IP address and show no errors.

Figure 13: Instances Window with Successfully Created Instance "Test1"

Extreme Networks OpenStack Plugin 2.0 Installation Guide

48

Managing Tenants and Virtual Machines

11 To see additional details about the instance, click the instance name. The Instance Detailwindow appears.

Figure 14: Instance Detail Window

Extreme Networks OpenStack Plugin 2.0 Installation Guide

49

Managing Tenants and Virtual Machines

12 Click the Console tab. The Instance Console window appears.

Figure 15: Instance Console Window Note If the screen does not respond, click the gray status bar. The VM has an IP address within the T1-net and can use the T1-virtual router and its VLANs to access Internet.

Migrating Tenant Virtual Machine Instances (Live Migration) A running VM instance can be migrated to a new server without any interruption to the users. The migration command can only be done from the controller using the nova API. Note VM migration is only supported between servers with compatible chipsets only. The following procedure below explains how to migrate the tenant1 VM instance t1Server1 created earlier from OSHost1 to OSHost2 (see Creating Tenant Virtual Machine Instances on page 44). 1

Authenticate as a user “admin” of tenant “tenant1”.

stack@OSController$ cd /opt/stack/devstack stack@OSController$ source openrc admin tenant1 2 List all VM instances of this tenant to make sure the intended instance exists. stack@OSController$ nova list 3 Perform the live migration. OSController> nova live-migration t1Server1 OSHost2

Extreme Networks OpenStack Plugin 2.0 Installation Guide

50

Managing Tenants and Virtual Machines

4 Confirm that the instance is correctly migrated to the new target server. See the following figure. +--------------------------+-----------+-------+----------------------+ | ID | Name | Status | Networks | +--------------------------+-----------+-------+----------------------+ | 753b2144-5d59-4670-. . . | t1Server1 | ACTIVE | tenant1net=10.0.1.2 | | b0daa302-b0b6-4ee3-. . . | t1Server2 | ACTIVE | tenant1net=10.0.1.3 | +--------------------------+-----------+-------+----------------------+

Figure 16: All Instances Window

Deleting Tenant Virtual Machine Instances To delete a tenant virtual machine (VM) instance: 1

Delete the interface using quantum router-interface-delete router_id port=port_id:

stack@OSController:~$ quantum router-interface-delete aa959db9-2f89-4bee-b99c-f05cbe41070b port=03a3d251-aebb-47f2-a53ada209fac8844 Removed interface from router aa959db9-2f89-4bee-b99c-f05cbe41070b. 2 Repeat preceding step until all ports belonging to the tenant subnet are detached from the router.

Extreme Networks OpenStack Plugin 2.0 Installation Guide

51

Managing Tenants and Virtual Machines

3 Use quantum router-port-list to check which existing ports are attached to the router: stack@OSController:~$ quantum router-port-list aa959db9-2f89-4beeb99c-f05cbe41070b +--------------------------------------+----------------------------+------------------+--------------------------------------------------------------------------------+ | id | name | mac_address | fixed_ips | +--------------------------------------+----------------------------+------------------+--------------------------------------------------------------------------------+ | 03a3d251-aebb-47f2-a53a-da209fac8844 | TOR2 VLAN port | fa:16:3e:2d:c4:b7 | {"subnet_id": "a24edd2f-16f2-4345a75f-34f5576b7754", "ip_address": "10.1.0.4"} | | 102a7f01-0a9b-469a-bec5-5d7dd926ec40 | T2-bgw3-TOR1 | fa:16:3e:5d:9d:28 | {"subnet_id": "dab4a5c3-d121-437f-add9c9e3f5a1bac9", "ip_address": "50.1.0.1"} | | 3517f1d4-adb0-4daf-9577-3a855dbdf55c | T2-int-subnet gateway port | fa:16:3e:c4:31:dd | {"subnet_id": "a24edd2f-16f2-4345a75f-34f5576b7754", "ip_address": "10.1.0.1"} | | 3edd4354-8583-4b23-9a94-11feb8105793 | T2-bgw1-subnet gateway port | fa:16:3e:94:ee:a4 | {"subnet_id": "430a0f09-d50e-4642b41c-62452ec0983b", "ip_address": "30.1.0.1"} | | 4e5c2815-88b6-422a-920c-56bdc22609ca | TOR1 VLAN port | fa:16:3e:11:78:c3 | {"subnet_id": "a24edd2f-16f2-4345a75f-34f5576b7754", "ip_address": "10.1.0.3"} | | 6002a994-f454-4826-9fbe-1e5ab12d2f25 | T2-bgw3-TOR2 | fa:16:3e:47:84:26 | {"subnet_id": "dab4a5c3-d121-437f-add9c9e3f5a1bac9", "ip_address": "50.1.0.2"} | | d3f6a0d9-daa7-4d6c-b351-cdaf4913e93b | TOR1 VLAN port | fa:16:3e:00:8c:60 | {"subnet_id": "430a0f09-d50e-4642b41c-62452ec0983b", "ip_address": "30.1.0.2"} | +--------------------------------------+----------------------------+------------------+--------------------------------------------------------------------------------+ 4 Delete the subnet using quantum subnet-delete subnet_id: stack@OSController:~$ quantum subnet-delete 1637bc33-c18b-4722-92a3d8a9903081e7 Deleted subnet: 1637bc33-c18b-4722-92a3-d8a9903081e7 stack@OSController:~$ 5 Delete the network using quantum net-delete network_id: stack@OSController:~$ quantum net-delete d088bdfe-c97d-43d0a23f-16e413d35631 Deleted network: d088bdfe-c97d-43d0-a23f-16e413d35631

Extreme Networks OpenStack Plugin 2.0 Installation Guide

52

9 Resetting the Testbed To reset servers and switches in the testbed back to their original state: 1

On OSController:

/opt/stack/devstack/unstack.sh sudo /etc/init.d/nfs-kernel-server stop sudo sed -i '/stack/d' /etc/exports sudo sed -i '/instances/d' /etc/exports sudo rm -rf /opt/stack && sudo rm -rf /etc/nova && sudo rm -rf /etc/ quantum && sudo rm -rf /etc/keystone && sudo rm -rf /etc/glance sudo reboot 2 On OSHost1 and OSHost2: /opt/stack/devstack/unstack.sh sudo /etc/init.d/umountnfs.sh stop sudo sed -i '/openstack/d' /etc/fstab sudo sed -i '/OSController/d' /etc/fstab sudo rm -rf /opt/cinder && sudo rm -rf /opt/stack && sudo rm -rf /etc/ nova && sudo rm -rf /etc/quantum && sudo rm -rf /etc/keystone && sudo rm -rf /etc/glance 3 On TOR1 and TOR2 switches: These switches should have the default.xsf file, so they only need to be completely unconfigured: unconfig switch all

A Glossary Compute node Controller DHCP Hypervisor Instance KVM LAG Libvert MLAG Tenant VM VR

Compute node Compute nodes form the resource core of the OpenStack compute cloud, providing the processing, memory, network, and storage resources to run instances.

Controller The controller orchestrates the network configuration of nodes including IP addresses, VLANs, bridging, and manages routing for both public and private networks. The controller provides virtual networks to enable compute servers to interact with each other and with the public network. All machines must have a public and private network interface.

DHCP Dynamic Host Configuration Protocol. DHCP is used by servers on an IP network to allocate IP addresses to computers. DHCP automates the IP address configuration of a computer without a network administrator. IP addresses are typically selected from a range of assigned IP addresses stored in a database on the server and issued to a computer which requests a new IP address.

Hypervisor A hypervisor or virtual machine monitor (VMM) is a piece of computer software, firmware or hardware that creates and runs virtual machines.

Instance A running virtual machine, or a virtual machine that can be used like a hardware server.

Glossary

KVM Kernel-based virtual machine. KVM is a virtualization infrastructure for the Linux kernel that turns it into a hypervisor. See Hypervisor.

LAG Link aggregation group. LAG is an open standards 802.3ad solution to bundle ports together for multipath support to increase resiliency and redundancy. LAGs allow you to combine (aggregate) multiple network connections in parallel to increase throughput beyond what a single connection could sustain, and allows you to provide redundancy if a links fails.

Libvert Libvirt is an open source API, daemon, and management tool for managing platform virtualization. It can be used to manage virtualization technologies, such as Linux KVM. These APIs are widely used in the orchestration layer of hypervisors in the development of a cloud-based solution. See KVM and Hypervisor.

MLAG Multi-chassis link aggregation group. MLAG is an evolution of 802.3ad that allows the bundled ports to be distributed to two chassis uplinks for chassis-level redundancy. See LAG.

Tenant The OpenStack compute system (Nova) is designed to be used by many different cloud computing consumers or customers acting as tenants on a shared system, using role-based access assignments. Tenants are isolated resource containers forming the principal organizational structure within the compute service. Tenants consist of a separate VLAN, volumes, instances, images, keys, and users.

VM Virtual machine. A VM is a software-based emulation of a computer. Virtual machines act like they have the computer architecture and functions of a physical computer. Some VMs emulate a complete system platform with a full operating system—others are designed to only run a particular program.

VR Virtual router. A virtual router is a software-based routing framework that allows a host machine to act as a typical hardware router over a local area network. A virtual router can enable a computer/server to have the abilities of an actual physical router by performing network and packet routing functionality of a router.

Extreme Networks OpenStack Plugin 2.0 Installation Guide

55