1 .. This work is licensed under a Creative Commons Attribution 4.0 International License.
2 .. http://creativecommons.org/licenses/by/4.0
3 .. (c) Open Platform for NFV Project, Inc. and its contributors
9 This document describes how to install the Fraser release of
10 OPNFV when using Fuel as a deployment tool, covering its usage,
11 limitations, dependencies and required system resources.
12 This is an unified documentation for both x86_64 and aarch64
13 architectures. All information is common for both architectures
14 except when explicitly stated.
20 This document provides guidelines on how to install and
21 configure the Fraser release of OPNFV when using Fuel as a
22 deployment tool, including required software and hardware configurations.
24 Although the available installation options provide a high degree of
25 freedom in how the system is set up, including architecture, services
26 and features, etc., said permutations may not provide an OPNFV
27 compliant reference architecture. This document provides a
28 step-by-step guide that results in an OPNFV Fraser compliant
31 The audience of this document is assumed to have good knowledge of
32 networking and Unix/Linux administration.
38 Before starting the installation of the Fraser release of
39 OPNFV, using Fuel as a deployment tool, some planning must be
45 Prior to installation, a number of deployment specific parameters must be collected, those are:
47 #. Provider sub-net and gateway information
49 #. Provider VLAN information
51 #. Provider DNS addresses
53 #. Provider NTP addresses
55 #. Network overlay you plan to deploy (VLAN, VXLAN, FLAT)
57 #. How many nodes and what roles you want to deploy (Controllers, Storage, Computes)
59 #. Monitoring options you want to deploy (Ceilometer, Syslog, etc.).
61 #. Other options not covered in the document are available in the links above
64 This information will be needed for the configuration procedures
65 provided in this document.
67 =========================================
68 Hardware Requirements for Virtual Deploys
69 =========================================
71 The following minimum hardware requirements must be met for the virtual
72 installation of Fraser using Fuel:
74 +----------------------------+--------------------------------------------------------+
75 | **HW Aspect** | **Requirement** |
77 +============================+========================================================+
78 | **1 Jumpserver** | A physical node (also called Foundation Node) that |
79 | | will host a Salt Master VM and each of the VM nodes in |
80 | | the virtual deploy |
81 +----------------------------+--------------------------------------------------------+
82 | **CPU** | Minimum 1 socket with Virtualization support |
83 +----------------------------+--------------------------------------------------------+
84 | **RAM** | Minimum 32GB/server (Depending on VNF work load) |
85 +----------------------------+--------------------------------------------------------+
86 | **Disk** | Minimum 100GB (SSD or SCSI (15krpm) highly recommended)|
87 +----------------------------+--------------------------------------------------------+
90 ===========================================
91 Hardware Requirements for Baremetal Deploys
92 ===========================================
94 The following minimum hardware requirements must be met for the baremetal
95 installation of Fraser using Fuel:
97 +-------------------------+------------------------------------------------------+
98 | **HW Aspect** | **Requirement** |
100 +=========================+======================================================+
101 | **# of nodes** | Minimum 5 |
103 | | - 3 KVM servers which will run all the controller |
106 | | - 2 Compute nodes |
108 +-------------------------+------------------------------------------------------+
109 | **CPU** | Minimum 1 socket with Virtualization support |
110 +-------------------------+------------------------------------------------------+
111 | **RAM** | Minimum 16GB/server (Depending on VNF work load) |
112 +-------------------------+------------------------------------------------------+
113 | **Disk** | Minimum 256GB 10kRPM spinning disks |
114 +-------------------------+------------------------------------------------------+
115 | **Networks** | 4 VLANs (PUBLIC, MGMT, STORAGE, PRIVATE) - can be |
116 | | a mix of tagged/native |
118 | | 1 Un-Tagged VLAN for PXE Boot - ADMIN Network |
120 | | Note: These can be allocated to a single NIC - |
121 | | or spread out over multiple NICs |
122 +-------------------------+------------------------------------------------------+
123 | **1 Jumpserver** | A physical node (also called Foundation Node) that |
124 | | hosts the Salt Master and MaaS VMs |
125 +-------------------------+------------------------------------------------------+
126 | **Power management** | All targets need to have power management tools that |
127 | | allow rebooting the hardware and setting the boot |
128 | | order (e.g. IPMI) |
129 +-------------------------+------------------------------------------------------+
133 All nodes including the Jumpserver must have the same architecture (either x86_64 or aarch64).
137 For aarch64 deployments an UEFI compatible firmware with PXE support is needed (e.g. EDK2).
139 ===============================
140 Help with Hardware Requirements
141 ===============================
143 Calculate hardware requirements:
145 For information on compatible hardware types available for use,
146 please see `Fuel OpenStack Hardware Compatibility List <https://www.mirantis.com/software/hardware-compatibility/>`_
148 When choosing the hardware on which you will deploy your OpenStack
149 environment, you should think about:
151 - CPU -- Consider the number of virtual machines that you plan to deploy in your cloud environment and the CPUs per virtual machine.
153 - Memory -- Depends on the amount of RAM assigned per virtual machine and the controller node.
155 - Storage -- Depends on the local drive space per virtual machine, remote volumes that can be attached to a virtual machine, and object storage.
157 - Networking -- Depends on the Choose Network Topology, the network bandwidth per virtual machine, and network storage.
159 ================================================
160 Top of the Rack (TOR) Configuration Requirements
161 ================================================
163 The switching infrastructure provides connectivity for the OPNFV
164 infrastructure operations, tenant networks (East/West) and provider
165 connectivity (North/South); it also provides needed connectivity for
166 the Storage Area Network (SAN).
167 To avoid traffic congestion, it is strongly suggested that three
168 physically separated networks are used, that is: 1 physical network
169 for administration and control, one physical network for tenant private
170 and public networks, and one physical network for SAN.
171 The switching connectivity can (but does not need to) be fully redundant,
172 in such case it comprises a redundant 10GE switch pair for each of the
173 three physically separated networks.
175 The physical TOR switches are **not** automatically configured from
176 the Fuel OPNFV reference platform. All the networks involved in the OPNFV
177 infrastructure as well as the provider networks and the private tenant
178 VLANs needs to be manually configured.
180 Manual configuration of the Fraser hardware platform should
181 be carried out according to the `OPNFV Pharos Specification
182 <https://wiki.opnfv.org/display/pharos/Pharos+Specification>`_.
184 ============================
185 OPNFV Software Prerequisites
186 ============================
188 The Jumpserver node should be pre-provisioned with an operating system,
189 according to the Pharos specification. Relevant network bridges should
190 also be pre-configured (e.g. admin_br, mgmt_br, public_br).
192 - The admin bridge (admin_br) is mandatory for the baremetal nodes PXE booting during Fuel installation.
193 - The management bridge (mgmt_br) is required for testing suites (e.g. functest/yardstick), it is
194 suggested to pre-configure it for debugging purposes.
195 - The public bridge (public_br) is also nice to have for debugging purposes, but not mandatory.
197 The user running the deploy script on the Jumpserver should belong to ``sudo`` and ``libvirt`` groups,
198 and have passwordless sudo access.
200 The following example adds the groups to the user ``jenkins``
204 $ sudo usermod -aG sudo jenkins
205 $ sudo usermod -aG libvirt jenkins
212 %jenkins ALL=(ALL) NOPASSWD:ALL
214 The folder containing the temporary deploy artifacts (``/home/jenkins/tmpdir`` in the examples below)
215 needs to have mask 777 in order for libvirt to be able to use them.
219 $ mkdir -p -m 777 /home/jenkins/tmpdir
221 For an AArch64 Jumpserver, the ``libvirt`` minimum required version is 3.x, 3.5 or newer highly recommended.
222 While not mandatory, upgrading the kernel and QEMU on the Jumpserver is also highly recommended
223 (especially on AArch64 Jumpservers).
225 For CentOS 7.4 (AArch64), distro provided packages are already new enough.
226 For Ubuntu 16.04 (arm64), distro packages are too old and 3rd party repositories should be used.
227 For convenience, Armband provides a DEB repository holding all the required packages.
229 To add and enable the Armband repository on an Ubuntu 16.04 system,
230 create a new sources list file ``/apt/sources.list.d/armband.list`` with the following contents:
234 $ cat /etc/apt/sources.list.d/armband.list
235 //for OpenStack Queens release
236 deb http://linux.enea.com/mcp-repos/queens/xenial queens-armband main
240 Fuel@OPNFV has been validated by CI using the following distributions
241 installed on the Jumpserver:
243 - CentOS 7 (recommended by Pharos specification);
248 The install script expects ``libvirt`` to be already running on the Jumpserver.
249 In case ``libvirt`` packages are missing, the script will install them; but
250 depending on the OS distribution, the user might have to start the ``libvirtd``
251 service manually, then run the deploy script again. Therefore, it
252 is recommended to install libvirt-bin explicitly on the Jumpserver before the deployment.
256 It is also recommended to install the newer kernel on the Jumpserver before the deployment.
260 The install script will automatically install the rest of required distro package
261 dependencies on the Jumpserver, unless explicitly asked not to (via ``-P`` deploy arg).
262 This includes Python, QEMU, libvirt etc.
266 The install script will alter Jumpserver sysconf and disable ``net.bridge.bridge-nf-call``.
270 $ apt-get install linux-image-generic-hwe-16.04-edge libvirt-bin
273 ==========================================
274 OPNFV Software Installation and Deployment
275 ==========================================
277 This section describes the process of installing all the components needed to
278 deploy the full OPNFV reference platform stack across a server cluster.
280 The installation is done with Mirantis Cloud Platform (MCP), which is based on
281 a reclass model. This model provides the formula inputs to Salt, to make the deploy
282 automatic based on deployment scenario.
283 The reclass model covers:
285 - Infrastructure node definition: Salt Master node (cfg01) and MaaS node (mas01)
286 - OpenStack node definition: Controller nodes (ctl01, ctl02, ctl03) and Compute nodes (cmp001, cmp002)
287 - Infrastructure components to install (software packages, services etc.)
288 - OpenStack components and services (rabbitmq, galera etc.), as well as all configuration for them
291 Automatic Installation of a Virtual POD
292 =======================================
294 For virtual deploys all the targets are VMs on the Jumpserver. The deploy script will:
296 - Create a Salt Master VM on the Jumpserver which will drive the installation
297 - Create the bridges for networking with virsh (only if a real bridge does not already exist for a given network)
298 - Install OpenStack on the targets
299 - Leverage Salt to install & configure OpenStack services
301 .. figure:: img/fuel_virtual.png
303 :alt: Fuel@OPNFV Virtual POD Network Layout Examples
305 Fuel@OPNFV Virtual POD Network Layout Examples
307 +-----------------------+------------------------------------------------------------------------+
308 | cfg01 | Salt Master VM |
309 +-----------------------+------------------------------------------------------------------------+
310 | ctl01 | Controller VM |
311 +-----------------------+------------------------------------------------------------------------+
312 | cmp001/cmp002 | Compute VMs |
313 +-----------------------+------------------------------------------------------------------------+
314 | gtw01 | Gateway VM with neutron services (dhcp agent, L3 agent, metadata, etc) |
315 +-----------------------+------------------------------------------------------------------------+
316 | odl01 | VM on which ODL runs (for scenarios deployed with ODL) |
317 +-----------------------+------------------------------------------------------------------------+
320 In this figure there are examples of two virtual deploys:
321 - Jumphost 1 has only virsh bridges, created by the deploy script
322 - Jumphost 2 has a mix of Linux and virsh bridges; When Linux bridge exists for a specified network,
323 the deploy script will skip creating a virsh bridge for it
327 A virtual network ``mcpcontrol`` is always created for initial connection of the VMs on Jumphost.
330 Automatic Installation of a Baremetal POD
331 =========================================
333 The baremetal installation process can be done by editing the information about
334 hardware and environment in the reclass files, or by using the files Pod Descriptor
335 File (PDF) and Installer Descriptor File (IDF) as described in the OPNFV Pharos project.
336 These files contain all the information about the hardware and network of the deployment
337 that will be fed to the reclass model during deployment.
339 The installation is done automatically with the deploy script, which will:
341 - Create a Salt Master VM on the Jumpserver which will drive the installation
342 - Create a MaaS Node VM on the Jumpserver which will provision the targets
343 - Install OpenStack on the targets
344 - Leverage MaaS to provision baremetal nodes with the operating system
345 - Leverage Salt to configure the operating system on the baremetal nodes
346 - Leverage Salt to install & configure OpenStack services
348 .. figure:: img/fuel_baremetal.png
350 :alt: Fuel@OPNFV Baremetal POD Network Layout Example
352 Fuel@OPNFV Baremetal POD Network Layout Example
354 +-----------------------+---------------------------------------------------------+
355 | cfg01 | Salt Master VM |
356 +-----------------------+---------------------------------------------------------+
357 | mas01 | MaaS Node VM |
358 +-----------------------+---------------------------------------------------------+
359 | kvm01..03 | Baremetals which hold the VMs with controller functions |
360 +-----------------------+---------------------------------------------------------+
361 | cmp001/cmp002 | Baremetal compute nodes |
362 +-----------------------+---------------------------------------------------------+
363 | prx01/prx02 | Proxy VMs for Nginx |
364 +-----------------------+---------------------------------------------------------+
365 | msg01..03 | RabbitMQ Service VMs |
366 +-----------------------+---------------------------------------------------------+
367 | dbs01..03 | MySQL service VMs |
368 +-----------------------+---------------------------------------------------------+
369 | mdb01..03 | Telemetry VMs |
370 +-----------------------+---------------------------------------------------------+
371 | odl01 | VM on which ODL runs (for scenarios deployed with ODL) |
372 +-----------------------+---------------------------------------------------------+
373 | Tenant VM | VM running in the cloud |
374 +-----------------------+---------------------------------------------------------+
376 In the baremetal deploy all bridges but "mcpcontrol" are Linux bridges. For the Jumpserver, it is
377 required to pre-configure at least the admin_br bridge for the PXE/Admin.
378 For the targets, the bridges are created by the deploy script.
382 A virtual network ``mcpcontrol`` is always created for initial connection of the VMs on Jumphost.
385 Steps to Start the Automatic Deploy
386 ===================================
388 These steps are common both for virtual and baremetal deploys.
390 #. Clone the Fuel code from gerrit
396 $ git clone https://git.opnfv.org/fuel
403 $ git clone https://git.opnfv.org/armband
406 #. Checkout the Fraser release
410 $ git checkout opnfv-6.2.1
412 #. Start the deploy script
414 Besides the basic options, there are other recommended deploy arguments:
416 - use ``-D`` option to enable the debug info
417 - use ``-S`` option to point to a tmp dir where the disk images are saved. The images will be
418 re-used between deploys
419 - use ``|& tee`` to save the deploy log to a file
423 $ ci/deploy.sh -l <lab_name> \
425 -b <URI to configuration repo containing the PDF file> \
428 -S <Storage directory for disk images> |& tee deploy.log
432 The deployment uses the OPNFV Pharos project as input (PDF and IDF files)
433 for hardware and network configuration of all current OPNFV PODs.
434 When deploying a new POD, one can pass the ``-b`` flag to the deploy script to override
435 the path for the labconfig directory structure containing the PDF and IDF (see below).
441 To start a virtual deployment, it is required to have the **virtual** keyword
442 while specifying the pod name to the installer script.
444 It will create the required bridges and networks, configure Salt Master and
449 $ ci/deploy.sh -l ericsson \
451 -s os-nosdn-nofeature-noha \
453 -S /home/jenkins/tmpdir |& tee deploy.log
455 Once the deployment is complete, the OpenStack Dashboard, Horizon, is
456 available at ``http://<controller VIP>:8078``
457 The administrator credentials are **admin** / **opnfv_secret**.
459 A simple (and generic) sample PDF/IDF set of configuration files may
460 be used for virtual deployments by setting lab/POD name to ``local-virtual1``.
461 This sample configuration is x86_64 specific and hardcodes certain parameters,
462 like public network address space, so a dedicated PDF/IDF is highly recommended.
466 $ ci/deploy.sh -l local \
468 -s os-nosdn-nofeature-noha \
470 -S /home/jenkins/tmpdir |& tee deploy.log
474 A x86 deploy on pod2 from Linux Foundation lab
478 $ ci/deploy.sh -l lf \
480 -s os-nosdn-nofeature-ha \
482 -S /home/jenkins/tmpdir |& tee deploy.log
484 .. figure:: img/lf_pod2.png
486 :alt: Fuel@OPNFV LF POD2 Network Layout
488 Fuel@OPNFV LF POD2 Network Layout
490 An aarch64 deploy on pod5 from Arm lab
494 $ ci/deploy.sh -l arm \
496 -s os-nosdn-nofeature-ha \
498 -S /home/jenkins/tmpdir |& tee deploy.log
500 .. figure:: img/arm_pod5.png
502 :alt: Fuel@OPNFV ARM POD5 Network Layout
504 Fuel@OPNFV ARM POD5 Network Layout
506 Once the deployment is complete, the SaltStack Deployment Documentation is
507 available at ``http://<proxy public VIP>:8090``.
509 When deploying a new POD, one can pass the ``-b`` flag to the deploy script to override
510 the path for the labconfig directory structure containing the PDF and IDF.
514 $ ci/deploy.sh -b file://<absolute_path_to_labconfig> \
519 -S <tmp_folder> |& tee deploy.log
521 - <absolute_path_to_labconfig> is the absolute path to a local directory, populated
522 similar to Pharos, i.e. PDF/IDF reside in ``<absolute_path_to_labconfig>/labs/<lab_name>``
523 - <lab_name> is the same as the directory in the path above
524 - <pod_name> is the name used for the PDF (``<pod_name>.yaml``) and IDF (``idf-<pod_name>.yaml``) files
528 Pod and Installer Descriptor Files
529 ==================================
531 Descriptor files provide the installer with an abstraction of the target pod
532 with all its hardware characteristics and required parameters. This information
533 is split into two different files:
534 Pod Descriptor File (PDF) and Installer Descriptor File (IDF).
536 The Pod Descriptor File is a hardware description of the pod
537 infrastructure. The information is modeled under a yaml structure.
538 A reference file with the expected yaml structure is available at
539 ``mcp/config/labs/local/pod1.yaml``.
541 The hardware description is arranged into a main "jumphost" node and a "nodes"
542 set for all target boards. For each node the following characteristics
545 - Node parameters including CPU features and total memory.
546 - A list of available disks.
547 - Remote management parameters.
548 - Network interfaces list including mac address, speed, advanced features and name.
552 The fixed IPs are ignored by the MCP installer script and it will instead
553 assign based on the network ranges defined in IDF.
555 The Installer Descriptor File extends the PDF with pod related parameters
556 required by the installer. This information may differ per each installer type
557 and it is not considered part of the pod infrastructure.
558 The IDF file must be named after the PDF with the prefix "idf-". A reference file with the expected
559 structure is available at ``mcp/config/labs/local/idf-pod1.yaml``.
561 The file follows a yaml structure and two sections "net_config" and "fuel" are expected.
563 The "net_config" section describes all the internal and provider networks
564 assigned to the pod. Each used network is expected to have a vlan tag, IP subnet and
565 attached interface on the boards. Untagged vlans shall be defined as "native".
567 The "fuel" section defines several sub-sections required by the Fuel installer:
569 - jumphost: List of bridge names for each network on the Jumpserver.
570 - network: List of device name and bus address info of all the target nodes.
571 The order must be aligned with the order defined in PDF file. Fuel installer relies on the IDF model
572 to setup all node NICs by defining the expected device name and bus address.
573 - maas: Defines the target nodes commission timeout and deploy timeout. (optional)
574 - reclass: Defines compute parameter tuning, including huge pages, cpu pinning
575 and other DPDK settings. (optional)
577 The following parameters can be defined in the IDF files under "reclass". Those value will
578 overwrite the default configuration values in Fuel repository:
580 - nova_cpu_pinning: List of CPU cores nova will be pinned to. Currently disabled.
581 - compute_hugepages_size: Size of each persistent huge pages. Usual values are '2M' and '1G'.
582 - compute_hugepages_count: Total number of persistent huge pages.
583 - compute_hugepages_mount: Mount point to use for huge pages.
584 - compute_kernel_isolcpu: List of certain CPU cores that are isolated from Linux scheduler.
585 - compute_dpdk_driver: Kernel module to provide userspace I/O support.
586 - compute_ovs_pmd_cpu_mask: Hexadecimal mask of CPUs to run DPDK Poll-mode drivers.
587 - compute_ovs_dpdk_socket_mem: Set of amount huge pages in MB to be used by OVS-DPDK daemon
588 taken for each NUMA node. Set size is equal to NUMA nodes count, elements are divided by comma.
589 - compute_ovs_dpdk_lcore_mask: Hexadecimal mask of DPDK lcore parameter used to run DPDK processes.
590 - compute_ovs_memory_channels: Number of memory channels to be used.
591 - dpdk0_driver: NIC driver to use for physical network interface.
592 - dpdk0_n_rxq: Number of RX queues.
595 The full description of the PDF and IDF file structure are available as yaml schemas.
596 The schemas are defined as a git submodule in Fuel repository. Input files provided
597 to the installer will be validated against the schemas.
599 - ``mcp/scripts/pharos/config/pdf/pod1.schema.yaml``
600 - ``mcp/scripts/pharos/config/pdf/idf-pod1.schema.yaml``
606 Please refer to the :ref:`Release Notes <fuel-release-notes-label>` article.
614 1) `OPNFV Home Page <https://www.opnfv.org>`_
615 2) `OPNFV documentation <https://docs.opnfv.org>`_
616 3) `Software downloads <https://www.opnfv.org/software/download>`_
620 4) `OpenStack Queens Release Artifacts <https://www.openstack.org/software/queens>`_
621 5) `OpenStack Documentation <https://docs.openstack.org>`_
625 6) `OpenDaylight Artifacts <https://www.opendaylight.org/software/downloads>`_
629 7) `Mirantis Cloud Platform Documentation <https://docs.mirantis.com/mcp/latest>`_
633 8) `Saltstack Documentation <https://docs.saltstack.com/en/latest/topics>`_
634 9) `Saltstack Formulas <https://salt-formulas.readthedocs.io/en/latest/develop/overview-reclass.html>`_
638 10) `Reclass model <https://reclass.pantsfullofunix.net>`_