Autogenerated update (2020-07-31)

Update:
- accessapproval_v1
- clouderrorreporting_v1beta1
- compute_alpha
- compute_beta
- compute_v1
- content_v2
- content_v2_1
- dataproc_v1
- dataproc_v1beta2
- doubleclickbidmanager_v1
- doubleclickbidmanager_v1_1
- drive_v2
- drive_v3
- firebasehosting_v1beta1
- monitoring_v1
- monitoring_v3
- realtimebidding_v1
- run_v1
- servicemanagement_v1
This commit is contained in:
Google APIs committed 2020-07-31 00:37:55 +00:00
1 parent 482f64818b
commit 72edce10c1
58 files changed
+4258 -1177

No files matched your search

+39 -14
View File
@@ -4155,7 +4155,8 @@ module Google
# The type of commitment, which affects the discount rate and the eligible
# resources. Type MEMORY_OPTIMIZED specifies a commitment that will only apply
# to memory optimized machines.
# to memory optimized machines. Type ACCELERATOR_OPTIMIZED specifies a
# commitment that will only apply to accelerator optimized machines.
# Corresponds to the JSON property `type`
# @return [String]
attr_accessor :type
@@ -8339,9 +8340,10 @@ module Google
# Google Compute Engine has two Health Check resources:
# * [Global](/compute/docs/reference/rest/`$api_version`/healthChecks) * [
# Regional](/compute/docs/reference/rest/`$api_version`/regionHealthChecks)
# Internal HTTP(S) load balancers use regional health checks. All other types of
# GCP load balancers and managed instance group auto-healing use global health
# checks. For more information, read Health Check Concepts.
# Internal HTTP(S) load balancers must use regional health checks. Internal TCP/
# UDP load balancers can use either regional or global health checks. All other
# types of GCP load balancers and managed instance group auto-healing must use
# global health checks. For more information, read Health Check Concepts.
# To perform health checks on network load balancers, you must use either
# httpHealthChecks or httpsHealthChecks.
class HealthCheck
@@ -12492,6 +12494,13 @@ module Google
class InstanceGroupManagersApplyUpdatesRequest
include Google::Apis::Core::Hashable
# Flag to update all instances instead of specified list of ?instances?. If the
# flag is set to true then the instances may not be specified in the request.
# Corresponds to the JSON property `allInstances`
# @return [Boolean]
attr_accessor :all_instances
alias_method :all_instances?, :all_instances
# The list of URLs of one or more instances for which you want to apply updates.
# Each URL can be a full URL or a partial URL, such as zones/[ZONE]/instances/[
# INSTANCE_NAME].
@@ -12529,6 +12538,7 @@ module Google
# Update properties of this object
def update!(**args)
@all_instances = args[:all_instances] if args.key?(:all_instances)
@instances = args[:instances] if args.key?(:instances)
@minimal_action = args[:minimal_action] if args.key?(:minimal_action)
@most_disruptive_allowed_action = args[:most_disruptive_allowed_action] if args.key?(:most_disruptive_allowed_action)
@@ -17614,6 +17624,7 @@ module Google
# @return [String]
attr_accessor :i_pv4_range
# Must be set to create a VPC network. If not set, a legacy network is created.
# When set to true, the VPC network is created in auto mode. When set to false,
# the VPC network is created in custom mode.
# An auto mode VPC network starts with one subnet per region. Each subnet has a
@@ -17765,7 +17776,8 @@ module Google
# reached, whether they are reachable, and where they are located. For more
# information about using NEGs, see Setting up internet NEGs or Setting up
# zonal NEGs. (== resource_for `$api_version`.networkEndpointGroups ==) (==
# resource_for `$api_version`.globalNetworkEndpointGroups ==)
# resource_for `$api_version`.globalNetworkEndpointGroups ==) (== resource_for `$
# api_version`.regionNetworkEndpointGroups ==)
class NetworkEndpointGroup
include Google::Apis::Core::Hashable
@@ -23784,6 +23796,13 @@ module Google
class RegionInstanceGroupManagersApplyUpdatesRequest
include Google::Apis::Core::Hashable
# Flag to update all instances instead of specified list of ?instances?. If the
# flag is set to true then the instances may not be specified in the request.
# Corresponds to the JSON property `allInstances`
# @return [Boolean]
attr_accessor :all_instances
alias_method :all_instances?, :all_instances
# The list of URLs of one or more instances for which you want to apply updates.
# Each URL can be a full URL or a partial URL, such as zones/[ZONE]/instances/[
# INSTANCE_NAME].
@@ -23821,6 +23840,7 @@ module Google
# Update properties of this object
def update!(**args)
@all_instances = args[:all_instances] if args.key?(:all_instances)
@instances = args[:instances] if args.key?(:instances)
@minimal_action = args[:minimal_action] if args.key?(:minimal_action)
@most_disruptive_allowed_action = args[:most_disruptive_allowed_action] if args.key?(:most_disruptive_allowed_action)
@@ -28319,8 +28339,9 @@ module Google
# @return [String]
attr_accessor :kind
# [Output Only] The position of the next byte of content from the serial console
# output. Use this value in the next request as the start parameter.
# [Output Only] The position of the next byte of content, regardless of whether
# the content exists, following the output returned in the `contents` property.
# Use this value in the next request as the start parameter.
# Corresponds to the JSON property `next`
# @return [Fixnum]
attr_accessor :next
@@ -28332,8 +28353,10 @@ module Google
# The starting byte position of the output that was returned. This should match
# the start parameter sent with the request. If the serial console output
# exceeds the size of the buffer, older output will be overwritten by newer
# content and the start values will be mismatched.
# exceeds the size of the buffer (1 MB), older output is overwritten by newer
# content. The output start value will indicate the byte position of the output
# that was returned, which might be different than the `start` value that was
# specified in the request.
# Corresponds to the JSON property `start`
# @return [Fixnum]
attr_accessor :start
@@ -30082,10 +30105,11 @@ module Google
attr_accessor :id
# The range of internal addresses that are owned by this subnetwork. Provide
# this property when you create the subnetwork. For example, 10.0.0.0/8 or 192.
# 168.0.0/16. Ranges must be unique and non-overlapping within a network. Only
# IPv4 is supported. This field is set at resource creation time. The range can
# be expanded after creation using expandIpCidrRange.
# this property when you create the subnetwork. For example, 10.0.0.0/8 or 100.
# 64.0.0/10. Ranges must be unique and non-overlapping within a network. Only
# IPv4 is supported. This field is set at resource creation time. This may be a
# RFC 1918 IP range, or a privately routed, non-RFC 1918 IP range, not belonging
# to Google. The range can be expanded after creation using expandIpCidrRange.
# Corresponds to the JSON property `ipCidrRange`
# @return [String]
attr_accessor :ip_cidr_range
@@ -30537,7 +30561,8 @@ module Google
# The range of IP addresses belonging to this subnetwork secondary range.
# Provide this property when you create the subnetwork. Ranges must be unique
# and non-overlapping with all primary and secondary IP ranges within a network.
# Only IPv4 is supported.
# Only IPv4 is supported. This may be a RFC 1918 IP range, or a privately, non-
# RFC 1918 IP range, not belonging to Google.
# Corresponds to the JSON property `ipCidrRange`
# @return [String]
attr_accessor :ip_cidr_range