Describes an autoscaling policy for Dataproc cluster autoscaler.
resource "google_dataproc_cluster" "basic" {
name = "dataproc-policy"
region = "us-central1"
cluster_config {
autoscaling_config {
policy_uri = google_dataproc_autoscaling_policy.asp.name
}
}
}
resource "google_dataproc_autoscaling_policy" "asp" {
policy_id = "dataproc-policy"
location = "us-central1"
worker_config {
max_instances = 3
}
basic_algorithm {
yarn_config {
graceful_decommission_timeout = "30s"
scale_up_factor = 0.5
scale_down_factor = 0.5
}
}
}
The following arguments are supported:
policy_id
-
(Required)
The policy id. The id must contain only letters (a-z, A-Z), numbers (0-9), underscores (_),
and hyphens (-). Cannot begin or end with underscore or hyphen. Must consist of between
3 and 50 characters.worker_config
-
(Optional)
Describes how the autoscaler will operate for primary workers.
Structure is documented below.
secondary_worker_config
-
(Optional)
Describes how the autoscaler will operate for secondary workers.
Structure is documented below.
basic_algorithm
-
(Optional)
Basic algorithm for autoscaling.
Structure is documented below.
location
-
(Optional)
The location where the autoscaling policy should reside.
The default value is global
.
project
- (Optional) The ID of the project in which the resource belongs.
If it is not provided, the provider project is used.
The worker_config
block supports:
min_instances
-
(Optional)
Minimum number of instances for this group. Bounds: [2, maxInstances]. Defaults to 2.
max_instances
-
(Required)
Maximum number of instances for this group.
weight
-
(Optional)
Weight for the instance group, which is used to determine the fraction of total workers
in the cluster from this instance group. For example, if primary workers have weight 2,
and secondary workers have weight 1, the cluster will have approximately 2 primary workers
for each secondary worker.
The cluster may not reach the specified balance if constrained by min/max bounds or other
autoscaling settings. For example, if maxInstances for secondary workers is 0, then only
primary workers will be added. The cluster can also be out of balance when created.
If weight is not set on any instance group, the cluster will default to equal weight for
all groups: the cluster will attempt to maintain an equal number of workers in each group
within the configured size bounds for each group. If weight is set for one group only,
the cluster will default to zero weight on the unset group. For example if weight is set
only on primary workers, the cluster will use primary workers only and no secondary workers.
The secondary_worker_config
block supports:
min_instances
-
(Optional)
Minimum number of instances for this group. Bounds: [0, maxInstances]. Defaults to 0.
max_instances
-
(Optional)
Maximum number of instances for this group. Note that by default, clusters will not use
secondary workers. Required for secondary workers if the minimum secondary instances is set.
Bounds: [minInstances, ). Defaults to 0.
weight
-
(Optional)
Weight for the instance group, which is used to determine the fraction of total workers
in the cluster from this instance group. For example, if primary workers have weight 2,
and secondary workers have weight 1, the cluster will have approximately 2 primary workers
for each secondary worker.
The cluster may not reach the specified balance if constrained by min/max bounds or other
autoscaling settings. For example, if maxInstances for secondary workers is 0, then only
primary workers will be added. The cluster can also be out of balance when created.
If weight is not set on any instance group, the cluster will default to equal weight for
all groups: the cluster will attempt to maintain an equal number of workers in each group
within the configured size bounds for each group. If weight is set for one group only,
the cluster will default to zero weight on the unset group. For example if weight is set
only on primary workers, the cluster will use primary workers only and no secondary workers.
The basic_algorithm
block supports:
cooldown_period
-
(Optional)
Duration between scaling events. A scaling period starts after the
update operation from the previous event has completed.
Bounds: [2m, 1d]. Default: 2m.
yarn_config
-
(Required)
YARN autoscaling configuration.
Structure is documented below.
The yarn_config
block supports:
graceful_decommission_timeout
-
(Required)
Timeout for YARN graceful decommissioning of Node Managers. Specifies the
duration to wait for jobs to complete before forcefully removing workers
(and potentially interrupting jobs). Only applicable to downscaling operations.
Bounds: [0s, 1d].
scale_up_factor
-
(Required)
Fraction of average pending memory in the last cooldown period for which to
add workers. A scale-up factor of 1.0 will result in scaling up so that there
is no pending memory remaining after the update (more aggressive scaling).
A scale-up factor closer to 0 will result in a smaller magnitude of scaling up
(less aggressive scaling).
Bounds: [0.0, 1.0].
scale_down_factor
-
(Required)
Fraction of average pending memory in the last cooldown period for which to
remove workers. A scale-down factor of 1 will result in scaling down so that there
is no available memory remaining after the update (more aggressive scaling).
A scale-down factor of 0 disables removing workers, which can be beneficial for
autoscaling a single job.
Bounds: [0.0, 1.0].
scale_up_min_worker_fraction
-
(Optional)
Minimum scale-up threshold as a fraction of total cluster size before scaling
occurs. For example, in a 20-worker cluster, a threshold of 0.1 means the autoscaler
must recommend at least a 2-worker scale-up for the cluster to scale. A threshold of
0 means the autoscaler will scale up on any recommended change.
Bounds: [0.0, 1.0]. Default: 0.0.
scale_down_min_worker_fraction
-
(Optional)
Minimum scale-down threshold as a fraction of total cluster size before scaling occurs.
For example, in a 20-worker cluster, a threshold of 0.1 means the autoscaler must
recommend at least a 2 worker scale-down for the cluster to scale. A threshold of 0
means the autoscaler will scale down on any recommended change.
Bounds: [0.0, 1.0]. Default: 0.0.
In addition to the arguments listed above, the following computed attributes are exported:
id
- an identifier for the resource with format projects/{{project}}/locations/{{location}}/autoscalingPolicies/{{policy_id}}
name
-
The "resource name" of the autoscaling policy.
This resource provides the following Timeouts configuration options:
create
- Default is 20 minutes.update
- Default is 20 minutes.delete
- Default is 20 minutes.AutoscalingPolicy can be imported using any of these accepted formats:
projects/{{project}}/locations/{{location}}/autoscalingPolicies/{{policy_id}}
{{project}}/{{location}}/{{policy_id}}
{{location}}/{{policy_id}}
In Terraform v1.5.0 and later, use an import
block to import AutoscalingPolicy using one of the formats above. For example:
import {
id = "projects/{{project}}/locations/{{location}}/autoscalingPolicies/{{policy_id}}"
to = google_dataproc_autoscaling_policy.default
}
When using the terraform import
command, AutoscalingPolicy can be imported using one of the formats above. For example:
$ terraform import google_dataproc_autoscaling_policy.default projects/{{project}}/locations/{{location}}/autoscalingPolicies/{{policy_id}}
$ terraform import google_dataproc_autoscaling_policy.default {{project}}/{{location}}/{{policy_id}}
$ terraform import google_dataproc_autoscaling_policy.default {{location}}/{{policy_id}}
This resource supports User Project Overrides.