Skip to main content
create_resource_group(name=None, tier_attributes=, ranking=None, adjoining_resource_group=”, options=)[source]

Creates a new resource group to facilitate resource management.

Parameters

name (str) –

Name of the group to be created. Must contain only letters, digits, and underscores, and cannot begin with a digit. Must not match existing resource group name.

tier_attributes (dict of str to dicts of str to str) –

Optional map containing tier names and their respective attribute group limits. The only valid attribute limit that can be set is max_memory (in bytes) for the VRAM and RAM tiers.

For instance, to set max VRAM capacity to 1GB per rank per GPU and max RAM capacity to 10GB per rank, use: ‘VRAM’:‘max_memory’:’1000000000’, ‘RAM’:‘max_memory’:’10000000000’. Allowed keys are:

  • max_memory – Maximum amount of memory usable at one time, per rank, per GPU, for the VRAM tier; or maximum amount of memory usable at one time, per rank, for the RAM tier.

The default value is an empty dict ( ).

ranking (str) –

Indicates the relative ranking among existing resource groups where this new resource group will be placed. Allowed values are:

  • first – Make this resource group the new first one in the ordering.

  • last – Make this resource group the new last one in the ordering.

  • before – Place this resource group before the one specified by input parameter adjoining_resource_group in the ordering.

  • after – Place this resource group after the one specified by input parameter adjoining_resource_group in the ordering.

adjoining_resource_group (str) –

If input parameter ranking is before or after, this field indicates the resource group before or after which the current group will be placed; otherwise, leave blank. The default value is ‘’.

options (dict of str to str) –

Optional parameters. Allowed keys are:

  • max_cpu_concurrency – Maximum number of simultaneous threads that will be used to execute a request, per rank, for this group. The minimum allowed value is ‘4’.

  • max_data – Maximum amount of data, per rank, in bytes, that can be used by all database objects within this group. Set to -1 to indicate no upper limit. The minimum allowed value is ‘-1’.

  • max_scheduling_priority – Maximum priority of a scheduled task for this group. The minimum allowed value is ‘1’. The maximum allowed value is ‘100’.

  • max_tier_priority – Maximum priority of a tiered object for this group. The minimum allowed value is ‘1’. The maximum allowed value is ‘10’.

The default value is an empty dict ( ).

Returns

A dict with the following entries–

name (str) –

Value of input parameter name.

info (dict of str to str) –

Additional information.