Cells

Before reading further, there is a nice overview presentation that Andrew Laski gave at the Austin (Newton) summit which is worth watching.

Cells V2

Manifesto

Proposal

Right now, when a request hits the Nova API for a particular instance, the instance information is fetched from the database, which contains the hostname of the compute node on which the instance currently lives. If the request needs to take action on the instance (which is most of them), the hostname is used to calculate the name of a queue, and a message is written there which finds its way to the proper compute node.

The meat of this proposal is changing the above hostname lookup into two parts that yield three pieces of information instead of one. Basically, instead of merely looking up the name of the compute node on which an instance lives, we will also obtain database and queue connection information. Thus, when asked to take action on instance $foo, we will:

  1. Lookup the three-tuple of (database, queue, hostname) for that instance

  2. Connect to that database and fetch the instance record

  3. Connect to the queue and send the message to the proper hostname queue

The above differs from the current organization in two ways. First, we need to do two database lookups before we know where the instance lives. Second, we need to demand-connect to the appropriate database and queue. Both of these have performance implications, but we believe we can mitigate the impacts through the use of things like a memcache of instance mapping information and pooling of connections to database and queue systems. The number of cells will always be much smaller than the number of instances.

There are availability implications with this change since something like a ‘nova list’ which might query multiple cells could end up with a partial result if there is a database failure in a cell. See CellsV2 Management for knowing more about the recommended practices under such situations. A database failure within a cell would cause larger issues than a partial list result so the expectation is that it would be addressed quickly and cellsv2 will handle it by indicating in the response that the data may not be complete.

Since this is very similar to what we have with current cells, in terms of organization of resources, we have decided to call this “cellsv2” for disambiguation.

After this work is complete there will no longer be a “no cells” deployment. The default installation of Nova will be a single cell setup.

Benefits

The benefits of this new organization are:

  • Native sharding of the database and queue as a first-class-feature in nova. All of the code paths will go through the lookup procedure and thus we won’t have the same feature parity issues as we do with current cells.

  • No high-level replication of all the cell databases at the top. The API will need a database of its own for things like the instance index, but it will not need to replicate all the data at the top level.

  • It draws a clear line between global and local data elements. Things like flavors and keypairs are clearly global concepts that need only live at the top level. Providing this separation allows compute nodes to become even more stateless and insulated from things like deleted/changed global data.

  • Existing non-cells users will suddenly gain the ability to spawn a new “cell” from their existing deployment without changing their architecture. Simply adding information about the new database and queue systems to the new index will allow them to consume those resources.

  • Existing cells users will need to fill out the cells mapping index, shutdown their existing cells synchronization service, and ultimately clean up their top level database. However, since the high-level organization is not substantially different, they will not have to re-architect their systems to move to cellsv2.

  • Adding new sets of hosts as a new “cell” allows them to be plugged into a deployment and tested before allowing builds to be scheduled to them.

Database split

As mentioned above there is a split between global data and data that is local to a cell.

The following is a breakdown of what data can uncontroversially considered global versus local to a cell. Missing data will be filled in as consensus is reached on the data that is more difficult to cleanly place. The missing data is mostly concerned with scheduling and networking.

Global (API-level) Tables

instance_types instance_type_projects instance_type_extra_specs quotas project_user_quotas quota_classes quota_usages security_groups security_group_rules security_group_default_rules provider_fw_rules key_pairs migrations networks tags

Cell-level Tables

instances instance_info_caches instance_extra instance_metadata instance_system_metadata instance_faults instance_actions instance_actions_events instance_id_mappings pci_devices block_device_mapping virtual_interfaces

Setup of Cells V2

Overview

As more of the CellsV2 implementation is finished, all operators are required to make changes to their deployment. For all deployments (even those that only intend to have one cell), these changes are configuration-related, both in the main nova configuration file as well as some extra records in the databases.

All nova deployments must now have the following databases available and configured:

  1. The “API” database

  2. One special “cell” database called “cell0”

  3. One (or eventually more) “cell” databases

Thus, a small nova deployment will have an API database, a cell0, and what we will call here a “cell1” database. High-level tracking information is kept in the API database. Instances that are never scheduled are relegated to the cell0 database, which is effectively a graveyard of instances that failed to start. All successful/running instances are stored in “cell1”.

Note

Since Nova services make use of both configuration file and some databases records, starting or restarting those services with an incomplete configuration could lead to an incorrect deployment. Please only restart the services once you are done with the described steps below.

First Time Setup

Since there is only one API database, the connection information for it is stored in the nova.conf file.

[api_database]
connection = mysql+pymysql://root:secretmysql@dbserver/nova_api?charset=utf8

Since there may be multiple “cell” databases (and in fact everyone will have cell0 and cell1 at a minimum), connection info for these is stored in the API database. Thus, you must have connection information in your config file for the API database before continuing to the steps below, so that nova-manage can find your other databases.

The following examples show the full expanded command line usage of the setup commands. This is to make it easier to visualize which of the various URLs are used by each of the commands. However, you should be able to put all of that in the config file and nova-manage will use those values. If need be, you can create separate config files and pass them as nova-manage –config-file foo.conf to control the behavior without specifying things on the command lines.

The commands below use the API database so remember to run nova-manage api_db sync first.

First we will create the necessary records for the cell0 database. To do that we use nova-manage like this:

nova-manage cell_v2 map_cell0 --database_connection \
  mysql+pymysql://root:secretmysql@dbserver/nova_cell0?charset=utf8

Note

If you don’t specify –database_connection then nova-manage will use the [database]/connection value from your config file, and mangle the database name to have a _cell0 suffix.

Warning

If your databases are on separate hosts then you should specify –database_connection or make certain that the nova.conf being used has the [database]/connection value pointing to the same user/password/host that will work for the cell0 database. If the cell0 mapping was created incorrectly, it can be deleted using the nova-manage cell_v2 delete_cell command and then run map_cell0 again with the proper database connection value.

Since no hosts are ever in cell0, nothing further is required for its setup. Note that all deployments only ever have one cell0, as it is special, so once you have done this step you never need to do it again, even if you add more regular cells.

Now, we must create another cell which will be our first “regular” cell, which has actual compute hosts in it, and to which instances can actually be scheduled. First, we create the cell record like this:

nova-manage cell_v2 create_cell --verbose --name cell1 \
  --database_connection  mysql+pymysql://root:secretmysql@127.0.0.1/nova?charset=utf8
  --transport-url rabbit://stackrabbit:secretrabbit@mqserver:5672/

Note

If you don’t specify the database and transport urls then nova-manage will use the [database]/connection and [DEFAULT]/transport_url values from the config file.

Note

At this point, the API database can now find the cell database, and further commands will attempt to look inside. If this is a completely fresh database (such as if you’re adding a cell, or if this is a new deployment), then you will need to run nova-manage db sync on it to initialize the schema.

The nova-manage cell_v2 create_cell command will print the UUID of the newly-created cell if –verbose is passed, which is useful if you need to run commands like discover_hosts targeted at a specific cell.

Now we have a cell, but no hosts are in it which means the scheduler will never actually place instances there. The next step is to scan the database for compute node records and add them into the cell we just created. For this step, you must have had a compute node started such that it registers itself as a running service. Once that has happened, you can scan and add it to the cell:

nova-manage cell_v2 discover_hosts

This command will connect to any databases for which you have created cells (as above), look for hosts that have registered themselves there, and map those hosts in the API database so that they are visible to the scheduler as available targets for instances. Any time you add more compute hosts to a cell, you need to re-run this command to map them from the top-level so they can be utilized.

Template URLs in Cell Mappings

Starting in the Rocky release, the URLs provided in the cell mappings for --database_connection and --transport-url can contain variables which are evaluated each time they are loaded from the database, and the values of which are taken from the corresponding base options in the host’s configuration file. The base URL is parsed and the following elements may be substituted into the cell mapping URL (using rabbit://bob:s3kret@myhost:123/nova?sync=true#extra):

Cell Mapping URL Variables

Variable

Meaning

Part of example URL

scheme

The part before the ://

rabbit

username

The username part of the credentials

bob

password

The password part of the credentials

s3kret

hostname

The hostname or address

myhost

port

The port number (must be specified)

123

path

The “path” part of the URL (without leading slash)

nova

query

The full query string arguments (without leading question mark)

sync=true

fragment

Everything after the first hash mark

extra

Variables are provided in curly brackets, like {username}. A simple template of rabbit://{username}:{password}@otherhost/{path} will generate a full URL of rabbit://bob:s3kret@otherhost/nova when used with the above example.

Note

The [database]/connection and [DEFAULT]/transport_url values are not reloaded from the configuration file during a SIGHUP, which means that a full service restart will be required to notice changes in a cell mapping record if variables are changed.

Note

The [DEFAULT]/transport_url option can contain an extended syntax for the “netloc” part of the url (i.e. userA:passwordA@hostA:portA,userB:passwordB:hostB:portB). In this case, substitions of the form username1, username2, etc will be honored and can be used in the template URL.

The templating of these URLs may be helpful in order to provide each service host with its own credentials for, say, the database. Without templating, all hosts will use the same URL (and thus credentials) for accessing services like the database and message queue. By using a URL with a template that results in the credentials being taken from the host-local configuration file, each host will use different values for those connections.

Assuming you have two service hosts that are normally configured with the cell0 database as their primary connection, their (abbreviated) configurations would look like this:

[database]
connection = mysql+pymysql://service1:foo@myapidbhost/nova_cell0

and:

[database]
connection = mysql+pymysql://service2:bar@myapidbhost/nova_cell0

Without cell mapping template URLs, they would still use the same credentials (as stored in the mapping) to connect to the cell databases. However, consider template URLs like the following:

mysql+pymysql://{username}:{password}@mycell1dbhost/nova

and:

mysql+pymysql://{username}:{password}@mycell2dbhost/nova

Using the first service and cell1 mapping, the calculated URL that will actually be used for connecting to that database will be:

mysql+pymysql://service1:foo@mycell1dbhost/nova

Step-By-Step for Common Use Cases

The following are step-by-step examples for common use cases setting up Cells V2. This is intended as a quick reference that puts together everything explained in Setup of Cells V2. It is assumed that you have followed all other install steps for Nova and are setting up Cells V2 specifically at this point.

Fresh Install

You are installing Nova for the first time and have no compute hosts in the database yet. This will set up a single cell Nova deployment.

  1. Reminder: You should have already created and synced the Nova API database by creating a database, configuring its connection in the [api_database]/connection setting in the Nova configuration file, and running nova-manage api_db sync.

  2. Create a database for cell0. If you are going to pass the database connection url on the command line in step 3, you can name the cell0 database whatever you want. If you are not going to pass the database url on the command line in step 3, you need to name the cell0 database based on the name of your existing Nova database: <Nova database name>_cell0. For example, if your Nova database is named nova, then your cell0 database should be named nova_cell0.

  3. Run the map_cell0 command to create and map cell0:

    nova-manage cell_v2 map_cell0 \
      --database_connection <database connection url>
    

    The database connection url is generated based on the [database]/connection setting in the Nova configuration file if not specified on the command line.

  4. Run nova-manage db sync to populate the cell0 database with a schema. The db sync command reads the database connection for cell0 that was created in step 3.

  5. Run the create_cell command to create the single cell which will contain your compute hosts:

    nova-manage cell_v2 create_cell --name <name> \
      --transport-url <transport url for message queue> \
      --database_connection <database connection url>
    

    The transport url is taken from the [DEFAULT]/transport_url setting in the Nova configuration file if not specified on the command line. The database url is taken from the [database]/connection setting in the Nova configuration file if not specified on the command line.

  6. Configure your compute host(s), making sure [DEFAULT]/transport_url matches the transport URL for the cell created in step 5, and start the nova-compute service. Before step 7, make sure you have compute hosts in the database by running:

    nova service-list --binary nova-compute
    
  7. Run the discover_hosts command to map compute hosts to the single cell by running:

    nova-manage cell_v2 discover_hosts
    

    The command will search for compute hosts in the database of the cell you created in step 5 and map them to the cell. You can also configure a periodic task to have Nova discover new hosts automatically by setting the [scheduler]/discover_hosts_in_cells_interval to a time interval in seconds. The periodic task is run by the nova-scheduler service, so you must be sure to configure it on all of your nova-scheduler hosts.

Note

Remember: In the future, whenever you add new compute hosts, you will need to run the discover_hosts command after starting them to map them to the cell if you did not configure the automatic host discovery in step 7.

Upgrade (minimal)

You are upgrading an existing Nova install and have compute hosts in the database. This will set up a single cell Nova deployment.

  1. If you haven’t already created a cell0 database in a prior release, create a database for cell0 with a name based on the name of your Nova database: <Nova database name>_cell0. If your Nova database is named nova, then your cell0 database should be named nova_cell0.

Warning

In Newton, the simple_cell_setup command expects the name of the cell0 database to be based on the name of the Nova API database: <Nova API database name>_cell0 and the database connection url is taken from the [api_database]/connection setting in the Nova configuration file.

  1. Run the simple_cell_setup command to create and map cell0, create and map the single cell, and map existing compute hosts and instances to the single cell:

    nova-manage cell_v2 simple_cell_setup \
      --transport-url <transport url for message queue>
    

    The transport url is taken from the [DEFAULT]/transport_url setting in the Nova configuration file if not specified on the command line. The database connection url will be generated based on the [database]/connection setting in the Nova configuration file.

Note

Remember: In the future, whenever you add new compute hosts, you will need to run the discover_hosts command after starting them to map them to the cell. You can also configure a periodic task to have Nova discover new hosts automatically by setting the [scheduler]/discover_hosts_in_cells_interval to a time interval in seconds. The periodic task is run by the nova-scheduler service, so you must be sure to configure it on all of your nova-scheduler hosts.

Upgrade with Cells V1

Todo

This needs to be removed but Adding a new cell to an existing deployment is still using it.

You are upgrading an existing Nova install that has Cells V1 enabled and have compute hosts in your databases. This will set up a multiple cell Nova deployment. At this time, it is recommended to keep Cells V1 enabled during and after the upgrade as multiple Cells V2 cell support is not fully finished and may not work properly in all scenarios. These upgrade steps will help ensure a simple cutover from Cells V1 to Cells V2 in the future.

Note

There is a Rocky summit video from CERN about how they did their upgrade from cells v1 to v2 here:

https://www.openstack.org/videos/vancouver-2018/moving-from-cellsv1-to-cellsv2-at-cern

  1. If you haven’t already created a cell0 database in a prior release, create a database for cell0. If you are going to pass the database connection url on the command line in step 2, you can name the cell0 database whatever you want. If you are not going to pass the database url on the command line in step 2, you need to name the cell0 database based on the name of your existing Nova database: <Nova database name>_cell0. For example, if your Nova database is named nova, then your cell0 database should be named nova_cell0.

  2. Run the map_cell0 command to create and map cell0:

    nova-manage cell_v2 map_cell0 \
      --database_connection <database connection url>
    

    The database connection url is generated based on the [database]/connection setting in the Nova configuration file if not specified on the command line.

  3. Run nova-manage db sync to populate the cell0 database with a schema. The db sync command reads the database connection for cell0 that was created in step 2.

  4. Run the create_cell command to create cells which will contain your compute hosts:

    nova-manage cell_v2 create_cell --name <cell name> \
      --transport-url <transport url for message queue> \
      --database_connection <database connection url>
    

    You will need to repeat this step for each cell in your deployment. Your existing cell database will be re-used – this simply informs the top-level API database about your existing cell databases.

    It is a good idea to specify a name for the new cell you create so you can easily look up cell uuids with the list_cells command later if needed.

    The transport url is taken from the [DEFAULT]/transport_url setting in the Nova configuration file if not specified on the command line. The database url is taken from the [database]/connection setting in the Nova configuration file if not specified on the command line. If you are not going to specify --database_connection and --transport-url on the command line, be sure to specify your existing cell Nova configuration file:

    nova-manage --config-file <cell nova.conf> cell_v2 create_cell \
      --name <cell name>
    
  5. Run the discover_hosts command to map compute hosts to cells:

    nova-manage cell_v2 discover_hosts --cell_uuid <cell uuid>
    

    You will need to repeat this step for each cell in your deployment unless you omit the --cell_uuid option. If the cell uuid is not specified on the command line, discover_hosts will search for compute hosts in each cell database and map them to the corresponding cell. You can use the list_cells command to look up cell uuids if you are going to specify --cell_uuid.

    You can also configure a periodic task to have Nova discover new hosts automatically by setting the [scheduler]/discover_hosts_in_cells_interval to a time interval in seconds. The periodic task is run by the nova-scheduler service, so you must be sure to configure it on all of your nova-scheduler hosts.

  6. Run the map_instances command to map instances to cells:

    nova-manage cell_v2 map_instances --cell_uuid <cell uuid> \
      --max-count <max count>
    

    You will need to repeat this step for each cell in your deployment. You can use the list_cells command to look up cell uuids.

    The --max-count option can be specified if you would like to limit the number of instances to map in a single run. If --max-count is not specified, all instances will be mapped. Repeated runs of the command will start from where the last run finished so it is not necessary to increase --max-count to finish. An exit code of 0 indicates that all instances have been mapped. An exit code of 1 indicates that there are remaining instances that need to be mapped.

Note

Remember: In the future, whenever you add new compute hosts, you will need to run the discover_hosts command after starting them to map them to a cell if you did not configure the automatic host discovery in step 5.

Adding a new cell to an existing deployment

To expand your deployment with a new cell, first follow the usual steps for standing up a new Cells V1 cell. After that is finished, follow step 4 in Upgrade with Cells V1 to create a new Cells V2 cell for it. If you have added new compute hosts for the new cell, you will also need to run the discover_hosts command after starting them to map them to the new cell if you did not configure the automatic host discovery as described in step 5 in Upgrade with Cells V1.

FAQs

  1. How do I find out which hosts are bound to which cell?

    There are a couple of ways to do this.

    1. Run nova-manage --config-file <cell config> host list. This will only lists hosts in the provided cell nova.conf.

      Deprecated since version 16.0.0: The nova-manage host list command is deprecated as of the 16.0.0 Pike release.

    2. Run nova-manage cell_v2 discover_hosts --verbose. This does not produce a report but if you are trying to determine if a host is in a cell you can run this and it will report any hosts that are not yet mapped to a cell and map them. This command is idempotent.

    3. Run nova-manage cell_v2 list_hosts. This will list hosts in all cells. If you want to list hosts in a specific cell, you can run nova-manage cell_v2 list_hosts --cell_uuid <cell_uuid>.

  2. I updated the database_connection and/or transport_url in a cell using the nova-manage cell_v2 update_cell command but the API is still trying to use the old settings.

    The cell mappings are cached in the nova-api service worker so you will need to restart the worker process to rebuild the cache. Note that there is another global cache tied to request contexts, which is used in the nova-conductor and nova-scheduler services, so you might need to do the same if you are having the same issue in those services. As of the 16.0.0 Pike release there is no timer on the cache or hook to refresh the cache using a SIGHUP to the service.

  3. I have upgraded from Newton to Ocata and I can list instances but I get a 404 NotFound error when I try to get details on a specific instance.

    Instances need to be mapped to cells so the API knows which cell an instance lives in. When upgrading, the nova-manage cell_v2 simple_cell_setup command will automatically map the instances to the single cell which is backed by the existing nova database. If you have already upgraded and did not use the simple_cell_setup command, you can run the nova-manage cell_v2 map_instances --cell_uuid <cell_uuid> command to map all instances in the given cell. See the Nova Cells v2 man page for details on command usage.

  4. Should I change any of the [cells] configuration options for Cells v2?

    NO. Those options are for Cells v1 usage only and are not used at all for Cells v2. That includes the nova-cells service - it has nothing to do with Cells v2.

  5. Can I create a cell but have it disabled from scheduling?

    Yes. It is possible to create a pre-disabled cell such that it does not become a candidate for scheduling new VMs. This can be done by running the nova-manage cell_v2 create_cell command with the --disabled option.

  6. How can I disable a cell so that the new server create requests do not go to it while I perform maintenance?

    Existing cells can be disabled by running nova-manage cell_v2 update_cell --cell_uuid <cell_uuid> --disable and can be re-enabled once the maintenance period is over by running nova-manage cell_v2 update_cell --cell_uuid <cell_uuid> --enable

  7. I disabled (or enabled) a cell using the nova-manage cell_v2 update_cell or I created a new (pre-disabled) cell(mapping) using the nova-manage cell_v2 create_cell command but the scheduler is still using the old settings.

    The cell mappings are cached in the scheduler worker so you will either need to restart the scheduler process to refresh the cache, or send a SIGHUP signal to the scheduler by which it will automatically refresh the cells cache and the changes will take effect.

  8. Why was the cells REST API not implemented for CellsV2? Why are there no CRUD operations for cells in the API?

    One of the deployment challenges that CellsV1 had was the requirement for the API and control services to be up before a new cell could be deployed. This was not a problem for large-scale public clouds that never shut down, but is not a reasonable requirement for smaller clouds that do offline upgrades and/or clouds which could be taken completely offline by something like a power outage. Initial devstack and gate testing for CellsV1 was delayed by the need to engineer a solution for bringing the services partially online in order to deploy the rest, and this continues to be a gap for other deployment tools. Consider also the FFU case where the control plane needs to be down for a multi-release upgrade window where changes to cell records have to be made. This would be quite a bit harder if the way those changes are made is via the API, which must remain down during the process.

    Further, there is a long-term goal to move cell configuration (i.e. cell_mappings and the associated URLs and credentials) into config and get away from the need to store and provision those things in the database. Obviously a CRUD interface in the API would prevent us from making that move.

  9. Why are cells not exposed as a grouping mechanism in the API for listing services, instances, and other resources?

    Early in the design of CellsV2 we set a goal to not let the cell concept leak out of the API, even for operators. Aggregates are the way nova supports grouping of hosts for a variety of reasons, and aggregates can cut across cells, and/or be aligned with them if desired. If we were to support cells as another grouping mechanism, we would likely end up having to implement many of the same features for them as aggregates, such as scheduler features, metadata, and other searching/filtering operations. Since aggregates are how Nova supports grouping, we expect operators to use aggregates any time they need to refer to a cell as a group of hosts from the API, and leave actual cells as a purely architectural detail.

    The need to filter instances by cell in the API can and should be solved by adding a generic by-aggregate filter, which would allow listing instances on hosts contained within any aggregate, including one that matches the cell boundaries if so desired.

  10. Why are the API responses for GET /servers, GET /servers/detail, GET /servers/{server_id} and GET /os-services missing some information for certain cells at certain times? Why do I see the status as “UNKNOWN” for the servers in those cells at those times when I run openstack server list or openstack server show?

    Starting from microversion 2.69 the API responses of GET /servers, GET /servers/detail, GET /servers/{server_id} and GET /os-services may contain missing keys during down cell situations. See the Handling Down Cells section of the Compute API guide for more information on the partial constructs.

    For administrative considerations, see Handling cell failures.