Deploy an additional nova cell v2

Deploy an additional nova cell v2

Warning

This currently is only supported in Stein or newer versions.

This guide assumes that you are ready to deploy a new overcloud, or have already installed an overcloud (min Stein release).

Note

Starting with CentOS 8 and TripleO Stein release, podman is the CONTAINERCLI to be used in the following steps.

Initial Deploy

The minimum requirement for having multiple cells is to have a central OpenStack controller cluster running all controller services. Additional cells will have cell controllers running the cell DB, cell MQ and a nova cell conductor service. In addition there are 1..n compute nodes. The central nova conductor service acts as a super conductor of the whole environment.

For more details on the cells v2 layout check Cells Layout (v2)

Note

Right now the current implementation does not support running nova metadata API per cell as explained in the cells v2 layout section Local per cell

The following example uses six nodes and the split control plane method to simulate a distributed cell deployment. The first Heat stack deploys a controller cluster and a compute. The second Heat stack deploys a cell controller and a compute node:

openstack overcloud status
+-----------+---------------------+---------------------+-------------------+
| Plan Name |       Created       |       Updated       | Deployment Status |
+-----------+---------------------+---------------------+-------------------+
| overcloud | 2019-02-12 09:00:27 | 2019-02-12 09:00:27 |   DEPLOY_SUCCESS  |
+-----------+---------------------+---------------------+-------------------+

openstack server list -c Name -c Status -c Networks
+----------------------------+--------+------------------------+
| Name                       | Status | Networks               |
+----------------------------+--------+------------------------+
| overcloud-controller-1     | ACTIVE | ctlplane=192.168.24.19 |
| overcloud-controller-2     | ACTIVE | ctlplane=192.168.24.11 |
| overcloud-controller-0     | ACTIVE | ctlplane=192.168.24.29 |
| overcloud-novacompute-0    | ACTIVE | ctlplane=192.168.24.15 |
+----------------------------+--------+------------------------+

The above deployed overcloud shows the nodes from the first stach and should be configured with Swift as Glance backend, using the following parameter, that images are pulled by the remote cell compute node over HTTP:

parameter_defaults:
  GlanceBackend: swift

Note

In this example the default cell and the additional cell uses the same network, When configuring another network scenario keep in mind that it will be necessary for the systems to be able to communicate with each other. E.g. the IPs for the default cell controller node will be in the endpoint map that later will be extracted from the overcloud stack and passed as a parameter to the second cell stack for it to access its endpoints. In this example both cells share an L2 network. In a production deployment it may be necessary instead to route.

Extract deployment information from the overcloud stack

Any additional cell stack requires information from the overcloud Heat stack where the central OpenStack services are located. The extracted parameters are needed as input for additional cell stacks. To extract these parameters into separate files in a directory (e.g. DIR=cell1) run the following:

source stackrc
mkdir cell1
export DIR=cell1
  1. Export the default cell EndpointMap

    openstack stack output show overcloud EndpointMap --format json \
    | jq '{"parameter_defaults": {"EndpointMapOverride": .output_value}}' \
    > $DIR/endpoint-map.json
    
  2. Export the default cell HostsEntry

    openstack stack output show overcloud HostsEntry -f json \
    | jq -r '{"parameter_defaults":{"ExtraHostFileEntries": .output_value}}' \
    > $DIR/extra-host-file-entries.json
    
  3. Export AllNodesConfig and GlobalConfig information

    In addition to the GlobalConfig, which contains the RPC information (port, ssl, scheme, user and password), additional information from the AllNodesConfig is required to point components to the default cell service instead of the service served by the cell controller. These are

    • oslo_messaging_notify_short_bootstrap_node_name - default cell overcloud messaging notify bootstrap node information

    • oslo_messaging_notify_node_names - default cell overcloud messaging notify node information

    • oslo_messaging_rpc_node_names - default cell overcloud messaging rpc node information as e.g. neutron agent needs to point to the overcloud messaging cluster

    • memcached_node_ips - memcached node information used by the cell services.

    (openstack stack output show overcloud AllNodesConfig --format json \
    | jq '.output_value | {oslo_messaging_notify_short_bootstrap_node_name: \
    .oslo_messaging_notify_short_bootstrap_node_name, \
    oslo_messaging_notify_node_names: .oslo_messaging_notify_node_names, \
    oslo_messaging_rpc_node_names: .oslo_messaging_rpc_node_names, \
    memcached_node_ips: .memcached_node_ips}'; \
    openstack stack output show overcloud GlobalConfig --format json \
    | jq '.output_value') |jq -s '.[0] * .[1]| {"parameter_defaults": \
    {"AllNodesExtraMapData": .}}' > $DIR/all-nodes-extra-map-data.json
    

    An example of a all-nodes-extra-map-data.json file:

    {
      "parameter_defaults": {
        "AllNodesExtraMapData": {
          "oslo_messaging_notify_short_bootstrap_node_name": "overcloud-controller-0",
          "oslo_messaging_notify_node_names": [
            "overcloud-controller-0.internalapi.site1.test",
            "overcloud-controller-1.internalapi.site1.test",
            "overcloud-controller-2.internalapi.site1.test"
          ],
          "oslo_messaging_rpc_node_names": [
            "overcloud-controller-0.internalapi.site1.test",
            "overcloud-controller-1.internalapi.site1.test",
            "overcloud-controller-2.internalapi.site1.test"
          ],
          "memcached_node_ips": [
            "172.16.2.232",
            "172.16.2.29",
            "172.16.2.49"
          ],
          "oslo_messaging_rpc_port": 5672,
          "oslo_messaging_rpc_use_ssl": "False",
          "oslo_messaging_notify_scheme": "rabbit",
          "oslo_messaging_notify_use_ssl": "False",
          "oslo_messaging_rpc_scheme": "rabbit",
          "oslo_messaging_rpc_password": "7l4lfamjPp6nqJgBMqb1YyM2I",
          "oslo_messaging_notify_password": "7l4lfamjPp6nqJgBMqb1YyM2I",
          "oslo_messaging_rpc_user_name": "guest",
          "oslo_messaging_notify_port": 5672,
          "oslo_messaging_notify_user_name": "guest"
        }
      }
    }
    
  4. Export passwords

    openstack object save --file - overcloud plan-environment.yaml \
    | python -c 'import yaml as y, sys as s; \
    s.stdout.write(y.dump({"parameter_defaults": \
    y.load(s.stdin.read())["passwords"]}));' > $DIR/passwords.yaml
    

    The same passwords are used for the cell services.

  5. Create roles file for cell stack

    openstack overcloud roles generate --roles-path \
    /usr/share/openstack-tripleo-heat-templates/roles \
    -o $DIR/cell_roles_data.yaml Compute CellController
    

    Note

    In case a different default heat stack name or compute role name is used, modify the above commands.

  6. Create cell parameter file for additional customization (e.g. cell1/cell1.yaml)

    Add the following content into a parameter file for the cell, e.g. cell1/cell1.yaml:

    resource_registry:
      # since the same network is used, the creation of the
      # different kind of networks is omitted for additional
      # cells
      OS::TripleO::Network::External: OS::Heat::None
      OS::TripleO::Network::InternalApi: OS::Heat::None
      OS::TripleO::Network::Storage: OS::Heat::None
      OS::TripleO::Network::StorageMgmt: OS::Heat::None
      OS::TripleO::Network::Tenant: OS::Heat::None
      OS::TripleO::Network::Management: OS::Heat::None
    
    parameter_defaults:
      # new CELL Parameter to reflect that this is an additional CELL
      NovaAdditionalCell: True
    
      # In case of an tls-everywhere environment the CloudName*
      # parameters need to be set for the cell as connection to
      # endpoints are done via DNS names, like MySQL and MQ  endpoints.
      # This is optional for non tls-everywhere environments.
      #CloudName: computecell1.ooo.test
      #CloudNameInternal: computecell1.internalapi.ooo.test
      #CloudNameStorage: computecell1.storage.ooo.test
      #CloudNameStorageManagement: computecell1.storagemgmt.ooo.test
      #CloudNameCtlplane: computecell1.ctlplane.ooo.test
    
      # CloudDomain is the same as in the default cell.
      #CloudDomain: ooo.test
    
      # Flavors used for the cell controller and computes
      OvercloudControllerFlavor: cellcontroller
      OvercloudComputeFlavor: compute
    
      # number of controllers/computes in the cell
      ControllerCount: 1
      ComputeCount: 1
      CephStorageCount: 0
    
      # default gateway
      ControlPlaneStaticRoutes:
        - ip_netmask: 0.0.0.0/0
          next_hop: 192.168.24.1
          default: true
      Debug: true
      DnsServers:
        - x.x.x.x
    

    The above file disables creating networks as the same as the overcloud stack created are used. It also specifies that this will be an additional cell using parameter NovaAdditionalCell.

Deploy the cell

  1. Create new flavor used to tag the cell controller

    openstack flavor create --id auto --ram 4096 --disk 40 --vcpus 1 cellcontroller
    openstack flavor set --property "cpu_arch"="x86_64" \
    --property "capabilities:boot_option"="local" \
    --property "capabilities:profile"="cellcontroller" \
    --property "resources:CUSTOM_BAREMETAL=1" \
    --property "resources:DISK_GB=0" \
    --property "resources:MEMORY_MB=0" \
    --property "resources:VCPU=0" \
    cellcontroller
    

    The properties need to be modified to the needs of the environment.

  2. Tag node into the new flavor using the following command

    openstack baremetal node set --property \
    capabilities='profile:cellcontroller,boot_option:local' <node id>
    

    Verify the tagged cellcontroller:

    openstack overcloud profiles list
    
  3. Deploy the cell

    To deploy the overcloud we can use use the same overcloud deploy command as it was used to deploy the overcloud stack and add the created export environment files:

    openstack overcloud deploy --override-ansible-cfg \
      /home/stack/custom_ansible.cfg \
      --stack computecell1 \
      --templates /usr/share/openstack-tripleo-heat-templates \
      -e /usr/share/openstack-tripleo-heat-templates/environments/docker-ha.yaml \
      -e ... additional environment files used for overcloud stack, like container
        prepare parameters, or other specific parameters for the cell
      ...
      -r $HOME/$DIR/cell_roles_data.yaml \
      -e $HOME/$DIR/passwords.yaml \
      -e $HOME/$DIR/endpoint-map.json \
      -e $HOME/$DIR/all-nodes-extra-map-data.json \
      -e $HOME/$DIR/extra-host-file-entries.json \
      -e $HOME/$DIR/cell1.yaml
    

    Wait for the deployment to finish:

    openstack stack list
    +--------------------------------------+--------------+----------------------------------+-----------------+----------------------+----------------------+
    | ID                                   | Stack Name   | Project                          | Stack Status    | Creation Time        | Updated Time         |
    +--------------------------------------+--------------+----------------------------------+-----------------+----------------------+----------------------+
    | 890e4764-1606-4dab-9c2f-6ed853e3fed8 | computecell1 | 2b303a97f4664a69ba2dbcfd723e76a4 | CREATE_COMPLETE | 2019-02-12T08:35:32Z | None                 |
    | 09531653-1074-4568-b50a-48a7b3cc15a6 | overcloud    | 2b303a97f4664a69ba2dbcfd723e76a4 | UPDATE_COMPLETE | 2019-02-09T09:52:56Z | 2019-02-11T08:33:37Z |
    +--------------------------------------+--------------+----------------------------------+-----------------+----------------------+----------------------+
    

Create the cell and discover compute nodes

  1. Add cell information to overcloud controllers

    On all central controllers add information on how to reach the messaging cell controller endpoint (usually internalapi) to /etc/hosts, from the undercloud:

    API_INFO=$(ssh heat-admin@<cell controlle ip> grep cellcontrol-0.internalapi /etc/hosts)
    ansible -i /usr/bin/tripleo-ansible-inventory Controller -b \
    -m lineinfile -a "dest=/etc/hosts line=\"$API_INFO\""
    

    Note

    Do this outside the HEAT_HOSTS_START .. HEAT_HOSTS_END block, or add it to an ExtraHostFileEntries section of an environment file for the central overcloud controller. Add the environment file to the next overcloud deploy run.

  2. Extract transport_url and database connection

    Get the transport_url and database connection endpoint information from the cell controller. This information is used to create the cell in the next step:

    ssh heat-admin@<cell controller ip> sudo crudini --get \
    /var/lib/config-data/nova/etc/nova/nova.conf DEFAULT transport_url
    ssh heat-admin@<cell controller ip> sudo crudini --get \
    /var/lib/config-data/nova/etc/nova/nova.conf database connection
    
  3. Create the cell

    Login to one of the central controllers create the cell with reference to the IP of the cell controller in the database_connection and the transport_url extracted from previous step, like:

    ssh heat-admin@<ctlplane ip overcloud-controller-0>
    
    # CONTAINERCLI can be either docker or podman
    export CONTAINERCLI='docker'
    
    sudo $CONTAINERCLI exec -it -u root nova_api /bin/bash
    nova-manage cell_v2 create_cell --name computecell1 \
    --database_connection \
    '{scheme}://{username}:{password}@172.16.2.102/nova?{query}' \
    --transport-url \
    'rabbit://guest:7l4lfamjPp6nqJgBMqb1YyM2I@computecell1-cellcontrol-0.internalapi.cell1.test:5672/?ssl=0'
    

    Note

    Templated transport cells URLs could be used if the same amount of controllers are in the default and add on cell.

    nova-manage cell_v2 list_cells --verbose
    

    After the cell got created the nova services on all central controllers need to be restarted.

    Docker:

    ansible -i /usr/bin/tripleo-ansible-inventory Controller -b -a \
    "docker restart nova_api nova_scheduler nova_conductor"
    

    Podman:

    ansible -i /usr/bin/tripleo-ansible-inventory Controller -b -a \
    "systemctl restart tripleo_nova_api tripleo_nova_conductor tripleo_nova_scheduler"
    
  4. Perform cell host discovery

    Login to one of the overcloud controllers and run the cell host discovery:

    ssh heat-admin@<ctlplane ip overcloud-controller-0>
    
    # CONTAINERCLI can be either docker or podman
    export CONTAINERCLI='docker'
    
    sudo $CONTAINERCLI exec -it -u root nova_api /bin/bash
    nova-manage cell_v2 discover_hosts --by-service --verbose
    nova-manage cell_v2 list_hosts
    
    +--------------+--------------------------------------+---------------------------------------+
    |  Cell Name   |              Cell UUID               |                Hostname               |
    +--------------+--------------------------------------+---------------------------------------+
    | computecell1 | 97bb4ee9-7fe9-4ec7-af0d-72b8ef843e3e | computecell1-novacompute-0.site1.test |
    |   default    | f012b67d-de96-471d-a44f-74e4a6783bca |   overcloud-novacompute-0.site1.test  |
    +--------------+--------------------------------------+---------------------------------------+
    

    The cell is now deployed and can be used.

Managing the cell

Add a compute to a cell

To increase resource capacity of a running cell, you can start more servers of a selected role. For more details on how to add nodes see Scaling overcloud roles.

After the node got deployed, login to one of the overcloud controllers and run the cell host discovery:

ssh heat-admin@<ctlplane ip overcloud-controller-0>

# CONTAINERCLI can be either docker or podman
export CONTAINERCLI='docker'

sudo $CONTAINERCLI exec -it -u root nova_api /bin/bash
nova-manage cell_v2 discover_hosts --by-service --verbose
nova-manage cell_v2 list_hosts

Delete a compute from a cell

As initial step migrate all instances off the compute.

  1. From one of the overcloud controllers, delete the computes from the cell:

    ssh heat-admin@<ctlplane ip overcloud-controller-0>
    
    # CONTAINERCLI can be either docker or podman
    export CONTAINERCLI='docker'
    
    sudo $CONTAINERCLI exec -it -u root nova_api /bin/bash
    nova-manage cell_v2 delete_host --cell_uuid <uuid> --host <compute>
    
  2. Delete the resource providers from placement

    This step is required as otherwise adding a compute node with the same hostname will make it to fail to register and update the resources with the placement service.:

    sudo yum install python2-osc-placement
    openstack resource provider list
    +--------------------------------------+---------------------------------------+------------+
    | uuid                                 | name                                  | generation |
    +--------------------------------------+---------------------------------------+------------+
    | 9cd04a8b-5e6c-428e-a643-397c9bebcc16 | computecell1-novacompute-0.site1.test |         11 |
    +--------------------------------------+---------------------------------------+------------+
    
    openstack resource provider delete 9cd04a8b-5e6c-428e-a643-397c9bebcc16
    
  3. Delete the node from the cell stack

Deleting a cell

As initial step delete all instances from cell

  1. From one of the overcloud controllers, delete all computes from the cell:

    ssh heat-admin@<ctlplane ip overcloud-controller-0>
    
    # CONTAINERCLI can be either docker or podman
    export CONTAINERCLI='docker'
    
    sudo $CONTAINERCLI exec -it -u root nova_api /bin/bash
    nova-manage cell_v2 delete_host --cell_uuid <uuid> --host <compute>
    
  2. On the cell controller delete all deleted instances:

    ssh heat-admin@<ctlplane ip cell controller>
    
    # CONTAINERCLI can be either docker or podman
    export CONTAINERCLI='docker'
    
    sudo $CONTAINERCLI exec -it -u root nova_conductor /bin/bash
    nova-manage db archive_deleted_rows --verbose
    
  3. From one of the overcloud controllers, delete the cell:

    ssh heat-admin@<ctlplane ip overcloud-controller-0>
    
    # CONTAINERCLI can be either docker or podman
    export CONTAINERCLI='docker'
    
    sudo $CONTAINERCLI exec -it -u root nova_api /bin/bash
    nova-manage cell_v2 delete_cell --cell_uuid <uuid>
    
  4. From a system which can reach the placement endpoint, delete the resource providers from placement

    This step is required as otherwise adding a compute node with the same hostname will make it to fail to register and update the resources with the placement service.:

    sudo yum install python2-osc-placement
    openstack resource provider list
    +--------------------------------------+---------------------------------------+------------+
    | uuid                                 | name                                  | generation |
    +--------------------------------------+---------------------------------------+------------+
    | 9cd04a8b-5e6c-428e-a643-397c9bebcc16 | computecell1-novacompute-0.site1.test |         11 |
    +--------------------------------------+---------------------------------------+------------+
    

    openstack resource provider delete 9cd04a8b-5e6c-428e-a643-397c9bebcc16

  5. Delete the cell stack:

    openstack stack delete computecell1 --wait --yes && openstack overcloud plan delete computecell1
    
Creative Commons Attribution 3.0 License

Except where otherwise noted, this document is licensed under Creative Commons Attribution 3.0 License. See all OpenStack Legal Documents.