Skip to content

tommycz1/salt-formula-nova

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Usage

OpenStack Nova provides a cloud computing fabric controller, supporting a wide variety of virtualization technologies, including KVM, Xen, LXC, VMware, and more. In addition to its native API, it includes compatibility with the commonly encountered Amazon EC2 and S3 APIs.

Sample Pillars

Controller nodes

Nova services on the controller node:

nova:
  controller:
    version: juno
    enabled: true
    security_group: true
    cpu_allocation_ratio: 8.0
    ram_allocation_ratio: 1.0
    disk_allocation_ratio: 1.0
    cross_az_attach: false
    workers: 8
    report_interval: 60
    dhcp_domain: novalocal
    consoleauth:
      token_ttl: 600
    bind:
      public_address: 10.0.0.122
      public_name: openstack.domain.com
      novncproxy_port: 6080
    database:
      engine: mysql
      host: 127.0.0.1
      port: 3306
      name: nova
      user: nova
      password: pwd
    identity:
      engine: keystone
      host: 127.0.0.1
      port: 35357
      user: nova
      password: pwd
      tenant: service
    message_queue:
      engine: rabbitmq
      host: 127.0.0.1
      port: 5672
      user: openstack
      password: pwd
      virtual_host: '/openstack'
    pci:
      alias:
        alias1:
          device_type: "type-PF"
          name: "a1"
          product_id: "154d"
          vendor_id: "8086"
    network:
      engine: neutron
      host: 127.0.0.1
      port: 9696
      extension_sync_interval: 600
      identity:
        engine: keystone
        host: 127.0.0.1
        port: 35357
        user: neutron
        password: pwd
        tenant: service
    metadata:
      password: password
    audit:
      enabled: false
    osapi_max_limit: 500
    barbican:
      enabled: true

Nova services from custom package repository:

nova:
  controller:
    version: juno
    source:
      engine: pkg
      address: http://...
  ....

Client-side RabbitMQ HA setup:

nova:
  controller:
    ....
    message_queue:
      engine: rabbitmq
      members:
        - host: 10.0.16.1
        - host: 10.0.16.2
        - host: 10.0.16.3
      user: openstack
      password: pwd
      virtual_host: '/openstack'
   ....

Enable auditing filter, i.e: CADF:

nova:
  controller:
    audit:
      enabled: true
  ....
      filter_factory: 'keystonemiddleware.audit:filter_factory'
      map_file: '/etc/pycadf/nova_api_audit_map.conf'
  ....

Enable CORS parameters:

nova:
  controller:
    cors:
      allowed_origin: https:localhost.local,http:localhost.local
      expose_headers: X-Auth-Token,X-Openstack-Request-Id,X-Subject-Token
      allow_methods: GET,PUT,POST,DELETE,PATCH
      allow_headers: X-Auth-Token,X-Openstack-Request-Id,X-Subject-Token
      allow_credentials: True
      max_age: 86400

Configuration of the policy.json file:

nova:
  controller:
    ....
    policy:
      context_is_admin: 'role:admin or role:administrator'
      'compute:create': 'rule:admin_or_owner'
      # Add key without value to remove line from policy.json
      'compute:create:attach_network':

Enable Barbican integration:

nova:
  controller:
    ....
    barbican:
      enabled: true

Define aliases for PCI devices: .. code-block:: yaml

nova:
controller:

... pci:

alias:
alias1:
device_type: "type-PF" name: "a1" product_id: "154d" vendor_id: "8086"

Enable cells update:

Note

Useful when upgrading Openstack. To update cells to test sync db agains duplicated production database.

nova:
  controller:
    update_cells: true

Configuring TLS communications

Note

By default system wide installed CA certs are used, so cacert_file param is optional, as well as cacert.

  • RabbitMQ TLS

    nova:
      compute:
         message_queue:
           port: 5671
           ssl:
             enabled: True
             (optional) cacert: cert body if the cacert_file does not exists
             (optional) cacert_file: /etc/openstack/rabbitmq-ca.pem
             (optional) version: TLSv1_2
  • MySQL TLS

    nova:
      controller:
         database:
           ssl:
             enabled: True
             (optional) cacert: cert body if the cacert_file does not exists
             (optional) cacert_file: /etc/openstack/mysql-ca.pem
  • Openstack HTTPS API

    Set the https as protocol at nova:compute and nova:controller sections :

    nova:
      controller :
         identity:
            protocol: https
            (optional) cacert_file: /etc/openstack/proxy.pem
          network:
            protocol: https
            (optional) cacert_file: /etc/openstack/proxy.pem
         glance:
            protocol: https
            (optional) cacert_file: /etc/openstack/proxy.pem
    nova:
      compute:
         identity:
            protocol: https
            (optional) cacert_file: /etc/openstack/proxy.pem
         network:
            protocol: https
            (optional) cacert_file: /etc/openstack/proxy.pem
         image:
            protocol: https
            (optional) cacert_file: /etc/openstack/proxy.pem
         ironic:
            protocol: https
            (optional) cacert_file: /etc/openstack/proxy.pem

Note

Barbican, Cinder, and placement url endpoints are discovering using service catalog.

Compute nodes

Nova controller services on compute node:

nova:
  compute:
    version: juno
    enabled: true
    cross_az_attach: false
    disk_cachemodes: network=writeback,block=none
    availability_zone: availability_zone_01
    aggregates:
    - hosts_with_fc
    - hosts_with_ssd
    security_group: true
    resume_guests_state_on_host_boot: False
    preallocate_images: space  # Default is 'none'
    my_ip: 10.1.0.16
    bind:
      vnc_address: 172.20.0.100
      vnc_port: 6080
      vnc_name: openstack.domain.com
      vnc_protocol: http
    database:
      engine: mysql
      host: 127.0.0.1
      port: 3306
      name: nova
      user: nova
      password: pwd
    identity:
      engine: keystone
      host: 127.0.0.1
      port: 35357
      user: nova
      password: pwd
      tenant: service
    message_queue:
      engine: rabbitmq
      host: 127.0.0.1
      port: 5672
      user: openstack
      password: pwd
      virtual_host: '/openstack'
    image:
      engine: glance
      host: 127.0.0.1
      port: 9292
    pci:
      alias:
        alias1:
          device_type: "type-PF"
          name: "a1"
          product_id: "154d"
          vendor_id: "8086"
    network:
      engine: neutron
      host: 127.0.0.1
      port: 9696
      identity:
        engine: keystone
        host: 127.0.0.1
        port: 35357
        user: neutron
        password: pwd
        tenant: service
    qemu:
      max_files: 4096
      max_processes: 4096
    host: node-12.domain.tld

Compute with vmware driver. Each vmware cluster requires a separate process of nova-compute. Each process should have uniq host identifier. However multiple computes might be running on single host. It is not recommended to have multiple computes running on different hosts that manage the same vmware cluster. To achive this pacemaker/corosync or keepalived might be used.

nova:
  compute:
    compute_driver: vmwareapi.VMwareVCDriver
    vmware:
      host_username: vmware
      host_password: vmware
      cluster_name: vmware_cluster01
      host_ip: 1.2.3.4

Group and user to be used for QEMU processes run by the system instance:

nova:
  compute:
    enabled: true
    ...
    qemu:
      user: nova
      group: cinder
      dynamic_ownership: 1

Group membership for user nova (upgrade related):

nova:
  compute:
    enabled: true
    ...
    user:
      groups:
      - libvirt

Nova services on compute node with OpenContrail:

nova:
  compute:
    enabled: true
    ...
    networking: contrail

Nova services on compute node with memcached caching and security strategy:

nova:
  compute:
    enabled: true
    ...
    cache:
      engine: memcached
      members:
      - host: 127.0.0.1
        port: 11211
      - host: 127.0.0.1
        port: 11211
      security:
        enabled: true
        strategy: ENCRYPT
        secret_key: secret

Client-side RabbitMQ HA setup:

nova:
  compute:
    ....
    message_queue:
      engine: rabbitmq
      members:
        - host: 10.0.16.1
        - host: 10.0.16.2
        - host: 10.0.16.3
      user: openstack
      password: pwd
      virtual_host: '/openstack'
   ....

Nova with ephemeral configured with Ceph:

nova:
  compute:
    enabled: true
    ...
    ceph:
      ephemeral: yes
      rbd_pool: nova
      rbd_user: nova
      secret_uuid: 03006edd-d957-40a3-ac4c-26cd254b3731
  ....

Nova with ephemeral configured with LVM:

nova:
  compute:
    enabled: true
    ...
    lvm:
      ephemeral: yes
      images_volume_group: nova_vg

linux:
  storage:
    lvm:
      nova_vg:
        name: nova_vg
        devices:
          - /dev/sdf
          - /dev/sdd
          - /dev/sdg
          - /dev/sde
          - /dev/sdc
          - /dev/sdj
          - /dev/sdh

Enable Barbican integration:

nova:
  compute:
    ....
    barbican:
      enabled: true

Define aliases for PCI devices: .. code-block:: yaml

nova:
compute:

... pci:

alias:
alias1:
device_type: "type-PF" name: "a1" product_id: "154d" vendor_id: "8086"

Nova metadata custom bindings:

nova:
  controller:
    enabled: true
    ...
    metadata:
      bind:
        address: 1.2.3.4
        port: 8776

Client role

Nova configured with NFS:

nova:
  compute:
    instances_path: /mnt/nova/instances

linux:
  storage:
    enabled: true
    mount:
      nfs_nova:
        enabled: true
        path: ${nova:compute:instances_path}
        device: 172.31.35.145:/data
        file_system: nfs
        opts: rw,vers=3

Nova flavors:

nova:
  client:
    enabled: true
    server:
      identity:
        flavor:
          flavor1:
            flavor_id: 10
            ram: 4096
            disk: 10
            vcpus: 1
          flavor2:
            flavor_id: auto
            ram: 4096
            disk: 20
            vcpus: 2
      identity1:
        flavor:
          ...

Availability zones:

nova:
  client:
    enabled: true
    server:
      identity:
        availability_zones:
        - availability_zone_01
        - availability_zone_02

Aggregates:

nova:
  client:
    enabled: true
    server:
      identity:
        aggregates:
        - aggregate1
        - aggregate2

Upgrade levels:

nova:
  controller:
    upgrade_levels:
      compute: juno

nova:
  compute:
    upgrade_levels:
      compute: juno

SR-IOV

Add PciPassthroughFilter into scheduler filters and NICs on specific compute nodes:

nova:
  controller:
    sriov: true
    scheduler_default_filters: "DifferentHostFilter,SameHostFilter,RetryFilter,AvailabilityZoneFilter,RamFilter,CoreFilter,DiskFilter,ComputeFilter,ComputeCapabilitiesFilter,ImagePropertiesFilter,ServerGroupAntiAffinityFilter,ServerGroupAffinityFilter,PciPassthroughFilter"

nova:
  compute:
    sriov:
      nic_one:
        devname: eth1
        physical_network: physnet1

Note

Parameters located under nova:compute:sriov:<nic_name> are copied to passthrough_whitelist parameter into nova.conf file in appropriate format.

CPU pinning & Hugepages

CPU pinning of virtual machine instances to dedicated physical CPU cores. Hugepages mount point for libvirt.

nova:
  controller:
    scheduler_default_filters: "DifferentHostFilter,SameHostFilter,RetryFilter,AvailabilityZoneFilter,RamFilter,CoreFilter,DiskFilter,ComputeFilter,ComputeCapabilitiesFilter,ImagePropertiesFilter,ServerGroupAntiAffinityFilter,ServerGroupAffinityFilter,NUMATopologyFilter,AggregateInstanceExtraSpecsFilter"

nova:
  compute:
    vcpu_pin_set: 2,3,4,5
    hugepages:
      mount_points:
      - path: /mnt/hugepages_1GB
      - path: /mnt/hugepages_2MB

Custom Scheduler filters

If you have a custom filter, that needs to be included in the scheduler, then you can include it like so:

nova:
  controller:
    scheduler_custom_filters:
    - my_custom_driver.nova.scheduler.filters.my_custom_filter.MyCustomFilter

    # Then add your custom filter on the end (make sure to include all other ones that you need as well)
    scheduler_default_filters: "DifferentHostFilter,SameHostFilter,RetryFilter,AvailabilityZoneFilter,RamFilter,CoreFilter,DiskFilter,ComputeFilter,ComputeCapabilitiesFilter,ImagePropertiesFilter,ServerGroupAntiAffinityFilter,ServerGroupAffinityFilter,PciPassthroughFilter,MyCustomFilter"

    # Since Queens version a sequence could be used as well:
    ~scheduler_default_filters:
       - DifferentHostFilter
       - SameHostFilter
         ...
       - MyCustomFilter

Hardware Trip/Unmap Support

To enable TRIM support for ephemeral images (thru nova managed images), libvirt has this option:

nova:
  compute:
    libvirt:
      hw_disk_discard: unmap

To actually utilize this feature, the following metadata must be set on the image as well, so the SCSI unmap is supported:

glance image-update --property hw_scsi_model=virtio-scsi <image>
glance image-update --property hw_disk_bus=scsi <image>

Scheduler Host Manager

Specify a custom host manager.

libvirt CPU mode

Allow setting the model of CPU that is exposed to a VM. This allows for better support live migration between hypervisors with different hardware, among other things. Defaults to host-passthrough.

nova:
  controller:
    scheduler_host_manager: ironic_host_manager

  compute:
    cpu_mode: host-model

Nova compute cpu model

nova:
  compute:
    cpu_mode: custom
    libvirt:
      cpu_model: IvyBridge

Nova compute workarounds

Live snapshotting is disabled by default in nova. To enable this, it needs a manual switch.

From manual:

When using libvirt 1.2.2 live snapshots fail intermittently under load
(likely related to concurrent libvirt/qemu operations). This config
option provides a mechanism to disable live snapshot, in favor of cold
snapshot, while this is resolved. Cold snapshot causes an instance
outage while the guest is going through the snapshotting process.

For more information, refer to the bug report:

https://bugs.launchpad.net/nova/+bug/1334398

Configurable pillar data:

nova:
  compute:
    workaround:
      disable_libvirt_livesnapshot: False

Config drive options

See example below on how to configure the options for the config drive:

nova:
  compute:
    config_drive:
      forced: True  # Default: True
      cdrom: True  # Default: False
      format: iso9660  # Default: vfat
      inject_password: False  # Default: False

Number of concurrent live migrates

Default is to have no concurrent live migrations (so 1 live-migration at a time).

Excerpt from config options page https://docs.openstack.org/ocata/config-reference/compute/config-options.html:

Maximum number of live migrations to run concurrently. This limit is enforced to avoid outbound live migrations overwhelming the host/network and causing failures. It is not recommended that you change this unless you are very sure that doing so is safe and stable in your environment.

Possible values:

  • 0 : treated as unlimited.
  • Negative value defaults to 0.
  • Any positive integer representing maximum number of live migrations to run concurrently.

To configure this option:

nova:
  compute:
    max_concurrent_live_migrations: 1  # (1 is the default)

Live migration with auto converge

Auto converge throttles down CPU if a progress of on-going live migration is slow https://docs.openstack.org/ocata/config-reference/compute/config-options.html:

nova:
  compute:
    libvirt:
      live_migration_permit_auto_converge: False  # (False is the default)
nova:
  controller:
    libvirt:
      live_migration_permit_auto_converge: False  # (False is the default)

Enhanced logging with logging.conf

By default logging.conf is disabled.

That is possible to enable per-binary logging.conf with new variables:

  • openstack_log_appender
    Set to true to enable log_config_append for all OpenStack services
  • openstack_fluentd_handler_enabled
    Set to true to enable FluentHandler for all Openstack services
  • openstack_ossyslog_handler_enabled
    Set to true to enable OSSysLogHandler for all Openstack services

Only WatchedFileHandler, OSSysLogHandler, and FluentHandler are available.

Also it is possible to configure this with pillar:

nova:
  controller:
      logging:
        log_appender: true
        log_handlers:
          watchedfile:
            enabled: true
          fluentd:
            enabled: true
          ossyslog:
            enabled: true

  compute:
      logging:
        log_appender: true
        log_handlers:
          watchedfile:
            enabled: true
          fluentd:
            enabled: true
          ossyslog:
            enabled: true

The log level might be configured per logger by using the following pillar structure:

nova:
  compute:
    logging:
      loggers:
        <logger_name>:
          level: WARNING

nova:
  compute:
    logging:
      loggers:
        <logger_name>:
          level: WARNING

Configure syslog parameters for libvirtd

To configure syslog parameters for libvirtd the below pillar structure should be used with values which are supported by libvirtd. These values might be known from the documentation.

nova:
  compute:
    libvirt:
      logging:
        level: 3
        filters: '3:remote 4:event'
        outputs: '3:syslog:libvirtd'
        buffer_size: 64

Logging controls:

Logging level: 4 errors, 3 warnings, 2 information, 1 debug basically 1 will log everything possible log_level = 3

Logging filters:

A filter allows to select a different logging level for a given category of logs.

The format for a filter is one of:

  • x:name
  • x:+name
    where name is a string which is matched against source file name, e.g., remote, qemu, or util/json, the optional + prefix tells libvirt to log stack trace for each message matching name, and x is the minimal level where matching messages should be logged:
  • 1: DEBUG
  • 2: INFO
  • 3: WARNING
  • 4: ERROR

Multiple filter can be defined in a single @filters, they just need to be separated by spaces.

For example, to only get warning or errors from the remote layer and only errors from the event layer: log_filters="3:remote 4:event

Logging outputs:

An output is one of the places to save logging information The format for an output can be:

  • x:stderr

    Output goes to stderr

  • x:syslog:name

    Use syslog for the output and use the given name as the ident

  • x:file:file_path

    output to a file, with the given filepath

    In all case the x prefix is the minimal level, acting as a filter

  • 1: DEBUG

  • 2: INFO

  • 3: WARNING

  • 4: ERROR

Multiple output can be defined, they just need to be separated by spaces. For example, to log all warnings and errors to syslog under the libvirt dident: log_outputs="3:syslog:libvirtd

Log debug buffer size: default 64 The daemon keeps an internal debug log buffer which will be dumped in case of crash or upon receiving a SIGUSR2 signal. This setting allows to override the default buffer size in kilobytes. If value is 0 or less the debug log buffer is deactivated log_buffer_size = 64

To configure the logging parameters for QEMU, the below pillar structure and logging parameters should be used:

nova:
  compute:
     qemu:
       logging:
         handler: logd
     virtlog:
       enabled: true
       level: 4
       filters: '3:remote 3:event'
       outputs: '4:syslog:virtlogd'
       max_clients: 512
       max_size: 2097100
       max_backups: 2

Inject password to VM

By default nova blocks up any inject to VM because inject_partition param is equal to -2. If you want to inject password to VM, you will need to define inject_partition greater or equal to -1 and define inject_password to True

For example:

nova:
  compute:
    inject_partition: '-1'
    inject_password: True

Allow the injection of an admin password for instance only at create and rebuild process.

There is no agent needed within the image to do this. If libguestfs is available on the host, it will be used. Otherwise nbd is used. The file system of the image will be mounted and the admin password, which is provided in the REST API call will be injected as password for the root user. If no root user is available, the instance won't be launched and an error is thrown. Be aware that the injection is not possible when the instance gets launched from a volume.

Possible values:

  • True
    Allows the injection
  • False (default)
    Disallows the injection. Any via the REST API provided admin password will be silently ignored.

Related options:

  • inject_partition
    Decides about the discovery and usage of the file system. It also can disable the injection at all. (boolean value)

You can read more about injecting the administrator password here: https://docs.openstack.org/nova/queens/admin/admin-password-injection.html

Enable libvirt control channel over TLS

By default TLS is disabled.

Enable TLS transport:

compute:
  libvirt:
    tls:
      enabled: True

You able to set custom certificates in pillar:

nova:
  compute:
    libvirt:
      tls:
        key: (certificate content)
        cert: (certificate content)
        cacert: (certificate content)
        client:
          key: (certificate content)
          cert: (certificate content)

Controlling access by tls_allowed_dn_list. Enable an access control list of client certificate Distinguished Names (DNs) which can connect to the TLS port on this server. The default is that DNs are not checked. This list may contain wildcards such as "C=GB,ST=London,L=London,O=Libvirt Project,CN=*" See the POSIX fnmatch function for the format of the wildcards. Note that if this is an empty list, no client can connect. Note also that GnuTLS returns DNs without spaces after commas between the fields (and this is what we check against), but the openssl x509 tool shows spaces.

nova:
  compute:
    libvirt:
      tls:
        tls_allowed_dn_list:
          host1:
            enabled: true
            value: 'C=foo,CN=cmp1'
          host2:
            enabled: true
            value: 'C=foo,CN=cmp2'

You can read more about live migration over TLS here: https://wiki.libvirt.org/page/TLSCreateServerCerts

Enable transport + authentication for VNC over TLS

# Only for Queens. Communication between noVNC proxy service and QEMU

By default communication between nova-novncproxy and qemu service is unsecure.

compute:
qemu:
vnc:
tls:
enabled: True
controller:
novncproxy:

# This section responsible for communication between noVNC proxy and client machine tls:

enabled: True

# This section responsible for communication between nova-novncproxy and qemu service vencrypt:

tls:
enabled: True

You able to set custom certificates in pillar:

nova:
compute:
qemu:
vnc:
tls:
cacert (certificate content) cert (certificate content) key (certificate content)
nova:
controller:
novncproxy:
tls:
server:
cert (certificate content) key (certificate content)
vencrypt:
tls:
cacert (certificate content) cert (certificate content) key (certificate content)
You can read more about it here:
https://docs.openstack.org/nova/queens/admin/remote-console-access.html

Enable communication between noVNC proxy and client machine over TLS

By default communication between noVNC proxy and client machine is unsecure.

controller:
novncproxy:
tls:
enabled: True
nova:
controller:
novncproxy:
tls:
server:
cert (certificate content) key (certificate content)
You can read more about it here:
https://docs.openstack.org/mitaka/config-reference/dashboard/configure.html

Enable x509 and ssl communication between Nova and Galera cluster.

By default communication between Nova and Galera is unsecure.

nova:
controller:
database:
x509:
enabled: True

You able to set custom certificates in pillar:

nova:
controller:
database:
x509:
cacert: (certificate content) cert: (certificate content) key: (certificate content)
You can read more about it here:
https://docs.openstack.org/security-guide/databases/database-access-control.html

Upgrades

Each openstack formula provide set of phases (logical bloks) that will help to build flexible upgrade orchestration logic for particular components. The list of phases might and theirs descriptions are listed in table below:

State Description
<app>.upgrade.service_running Ensure that all services for particular application are enabled for autostart and running
<app>.upgrade.service_stopped Ensure that all services for particular application disabled for autostart and dead
<app>.upgrade.pkgs_latest Ensure that packages used by particular application are installed to latest available version. This will not upgrade data plane packages like qemu and openvswitch as usually minimal required version in openstack services is really old. The data plane packages should be upgraded separately by apt-get upgrade or apt-get dist-upgrade Applying this state will not autostart service.
<app>.upgrade.render_config Ensure configuration is rendered actual version.
<app>.upgrade.pre We assume this state is applied on all nodes in the cloud before running upgrade. Only non destructive actions will be applied during this phase. Perform service built in service check like (keystone-manage doctor and nova-status upgrade)
<app>.upgrade.upgrade.pre Mostly applicable for data plane nodes. During this phase resources will be gracefully removed from current node if it is allowed. Services for upgraded application will be set to admin disabled state to make sure node will not participate in resources scheduling. For example on gtw nodes this will set all agents to admin disable state and will move all routers to other agents.
<app>.upgrade.upgrade This state will basically upgrade application on particular target. Stop services, render configuration, install new packages, run offline dbsync (for ctl), start services. Data plane should not be affected, only OpenStack python services.
<app>.upgrade.upgrade.post Add services back to scheduling.
<app>.upgrade.post This phase should be launched only when upgrade of the cloud is completed. Cleanup temporary files, perform other post upgrade tasks.
<app>.upgrade.verify Here we will do basic health checks (API CRUD operations, verify do not have dead network agents/compute services)

Documentation and Bugs

Packages

No packages published

Languages

  • Python 55.1%
  • SaltStack 29.1%
  • Scheme 5.0%
  • HTML 4.8%
  • Shell 3.4%
  • Makefile 2.6%