Quotas

Quotas

Nova uses a quota system for setting limits on resources such as number of instances or amount of CPU that a specific project or user can use.

Quotas are enforced by making a claim, or reservation, on resources when a request is made, such as creating a new server. If the claim fails, the request is rejected. If the reservation succeeds then the operation progresses until such a point that the reservation is either converted into usage (the operation was successful) or rolled back (the operation failed).

Typically the quota reservation is made in the nova-api service and the usage or rollback is performed in the nova-compute service, at least when dealing with a server creation or move operation.

Quota limits and usage can be retrieved via the limits REST API.

Checking quota

When calculating limits for a given resource and tenant, the following checks are made in order:

  • Depending on the resource, is there a tenant-specific limit on the resource in either the quotas or project_user_quotas tables in the database? If so, use that as the limit. You can create these resources by doing:

    openstack quota set --instances 5 <project>
    
  • Check to see if there is a hard limit for the given resource in the quota_classes table in the database for the default quota class. If so, use that as the limit. You can modify the default quota limit for a resource by doing:

    openstack quota set --class --instances 5 default
    
  • If the above does not provide a resource limit, then rely on the quota_* configuration options for the default limit.

Note

The API sets the limit in the quota_classes table. Once a default limit is set via the default quota class, that takes precedence over any changes to that resource limit in the configuration options. In other words, once you’ve changed things via the API, you either have to keep those synchronized with the configuration values or remove the default limit from the database manually as there is no REST API for removing quota class values from the database.

Quota usage from placement

Starting in the Train (20.0.0) release, it is possible to configure quota usage counting of cores and ram from the placement service and instances from instance mappings in the API database instead of counting resources from cell databases. This makes quota usage counting resilient in the presence of down or poor-performing cells.

Quota usage counting from placement is opt-in via configuration option:

[quota]
count_usage_from_placement = True

There are some things to note when opting in to counting quota usage from placement:

  • Counted usage will not be accurate in an environment where multiple Nova deployments are sharing a placement deployment because currently placement has no way of partitioning resource providers between different Nova deployments. Operators who are running multiple Nova deployments that share a placement deployment should not set the quota.count_usage_from_placement configuration option to True.

  • Behavior will be different for resizes. During a resize, resource allocations are held on both the source and destination (even on the same host, see https://bugs.launchpad.net/nova/+bug/1790204) until the resize is confirmed or reverted. Quota usage will be inflated for servers in this state and operators should weigh the advantages and disadvantages before enabling quota.count_usage_from_placement.

  • The populate_queued_for_delete and populate_user_id online data migrations must be completed before usage can be counted from placement. Until the data migration is complete, the system will fall back to legacy quota usage counting from cell databases depending on the result of an EXISTS database query during each quota check, if quota.count_usage_from_placement is set to True. Operators who want to avoid the performance hit from the EXISTS queries should wait to set the quota.count_usage_from_placement configuration option to True until after they have completed their online data migrations via nova-manage db online_data_migrations.

  • Behavior will be different for unscheduled servers in ERROR state. A server in ERROR state that has never been scheduled to a compute host will not have placement allocations, so it will not consume quota usage for cores and ram.

  • Behavior will be different for servers in SHELVED_OFFLOADED state. A server in SHELVED_OFFLOADED state will not have placement allocations, so it will not consume quota usage for cores and ram. Note that because of this, it will be possible for a request to unshelve a server to be rejected if the user does not have enough quota available to support the cores and ram needed by the server to be unshelved.

Known issues

If not counting quota usage from placement it is possible for down or poor performing cells to impact quota calculations. See the cells documentation for details.

Future plans

TODO: talk about quotas in the resource counting spec and nested quotas

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.