terraform/website/source/docs/providers/google/r/compute_autoscaler.html.mar...

136 lines
4.0 KiB
Markdown
Raw Normal View History

---
layout: "google"
page_title: "Google: google_compute_autoscaler"
sidebar_current: "docs-google-compute-autoscaler"
description: |-
Manages an Autoscaler within GCE.
---
# google\_compute\_autoscaler
A Compute Engine Autoscaler automatically adds or removes virtual machines from
a managed instance group based on increases or decreases in load. This allows
your applications to gracefully handle increases in traffic and reduces cost
when the need for resources is lower. You just define the autoscaling policy and
the autoscaler performs automatic scaling based on the measured load. For more
information, see [the official
documentation](https://cloud.google.com/compute/docs/autoscaler/) and
[API](https://cloud.google.com/compute/docs/autoscaler/v1beta2/autoscalers)
## Example Usage
```
resource "google_compute_instance_template" "foobar" {
name = "foobar"
machine_type = "n1-standard-1"
can_ip_forward = false
tags = ["foo", "bar"]
disk {
source_image = "debian-cloud/debian-7-wheezy-v20160301"
}
network_interface {
network = "default"
}
metadata {
foo = "bar"
}
service_account {
scopes = ["userinfo-email", "compute-ro", "storage-ro"]
}
}
resource "google_compute_target_pool" "foobar" {
name = "foobar"
}
resource "google_compute_instance_group_manager" "foobar" {
name = "foobar"
instance_template = "${google_compute_instance_template.foobar.self_link}"
target_pools = ["${google_compute_target_pool.foobar.self_link}"]
base_instance_name = "foobar"
zone = "us-central1-f"
}
resource "google_compute_autoscaler" "foobar" {
name = "foobar"
zone = "us-central1-f"
target = "${google_compute_instance_group_manager.foobar.self_link}"
autoscaling_policy = {
max_replicas = 5
min_replicas = 1
cooldown_period = 60
cpu_utilization = {
target = 0.5
}
}
}
```
2015-09-11 20:56:20 +02:00
## Argument Reference
The following arguments are supported:
* `description` - (Optional) An optional textual description of the instance
group manager.
* `target` - (Required) The full URL to the instance group manager whose size we
control.
* `autoscaling_policy.` - (Required) The parameters of the autoscaling
algorithm. Structure is documented below.
* `zone` - (Required) The zone of the target.
The `autoscaling_policy` block contains:
* `max_replicas` - (Required) The group will never be larger than this.
* `min_replicas` - (Required) The group will never be smaller than this.
* `cooldown_period` - (Optional) Period to wait between changes. This should be
at least double the time your instances take to start up.
* `cpu_utilization` - (Optional) A policy that scales when the cluster's average
CPU is above or below a given threshold. Structure is documented below.
* `metric` - (Optional) A policy that scales according to Google Cloud
Monitoring metrics Structure is documented below.
* `load_balancing_utilization` - (Optional) A policy that scales when the load
reaches a proportion of a limit defined in the HTTP load balancer. Structure
is documented below.
The `cpu_utilization` block contains:
* `target` - The floating point threshold where CPU utilization should be. E.g.
for 50% one would specify 0.5.
The `metric` block contains (more documentation
[here](https://cloud.google.com/monitoring/api/metrics)):
* `name` - The name of the Google Cloud Monitoring metric to follow, e.g.
`compute.googleapis.com/instance/network/received_bytes_count`
* `type` - Either "cumulative", "delta", or "gauge".
* `target` - The desired metric value per instance. Must be a positive value.
The `load_balancing_utilization` block contains:
* `target` - The floating point threshold where load balancing utilization
should be. E.g. if the load balancer's `maxRatePerInstance` is 10 requests
per second (RPS) then setting this to 0.5 would cause the group to be scaled
such that each instance receives 5 RPS.
## Attributes Reference
The following attributes are exported:
* `self_link` - The URL of the created resource.