$ gcloud compute instances create-with-container instance-1 \
    --zone us-central1-a             \
    --container-image=gcr.io/google-containers/busybox

POSITIONAL ARGUMENTS

INSTANCE_NAMES [INSTANCE_NAMES ...]

Names of the instances to create.

FLAGS

--boot-disk-auto-delete

Automatically delete boot disks when their instances are deleted. Enabled by default, use --no-boot-disk-auto-delete to disable.

--boot-disk-device-name=BOOT_DISK_DEVICE_NAME

The name the guest operating system will see for the boot disk. This option can only be specified if a new boot disk is being created (as opposed to mounting an existing persistent disk).

--boot-disk-size=BOOT_DISK_SIZE

The size of the boot disk. This option can only be specified if a new boot disk is being created (as opposed to mounting an existing persistent disk). The value must be a whole number followed by a size unit of KB for kilobyte, MB for megabyte, GB for gigabyte, or TB for terabyte. For example, 10GB will produce a 10 gigabyte disk. The minimum size a boot disk can have is 10 GB. Disk size must be a multiple of 1 GB. Limit your boot disk size to 2TB to account for MBR partition table limitations.

--boot-disk-type=BOOT_DISK_TYPE

The type of the boot disk. This option can only be specified if a new boot disk is being created (as opposed to mounting an existing persistent disk). To get a list of available disk types, run $ gcloud compute disk-types list.

--can-ip-forward

If provided, allows the instances to send and receive packets with non-matching destination or source IP addresses.

--container-arg=CONTAINER_ARG

Argument to append to container entrypoint or to override container CMD. Each argument must have a separate flag. Arguments are appended in the order of flags. Example:
Assuming the default entry point of your container (or an entry point overridden with --container-command flag) is a Bourne shell-compatible executable, in order to execute 'ls -l' command in the container, you could use:
--container-arg="-c" --container-arg="ls -l"
Caveat: due to the nature of the argument parsing, it's impossible to provide the flag value that starts with a dash (-) without the = sign (that is, --container-arg "-c" will not work correctly).
Default: None. (no arguments appended)

--container-command=CONTAINER_COMMAND

Specifies what executable to run when the container starts (overrides default entrypoint), eg. nc.
Default: None (default container entrypoint is used)

--container-env=[KEY=VALUE, ...,...]

Declare environment variables KEY with value VALUE passed to container. Only the last value of KEY is taken when KEY is repeated more than once.
Values, declared with --container-env flag override those with the same KEY from file, provided in --container-env-file.

--container-env-file=CONTAINER_ENV_FILE

Declare environment variables in a file. Values, declared with --container-env flag override those with the same KEY from file.
File with environment variables in format used by docker (almost). This means:

---

Lines are in format KEY=VALUE.

---

Values must contain equality signs.

---

Variables without values are not supported (this is different from docker format).

---

If # is first non-whitespace character in a line the line is ignored as a comment.

---

Lines with nothing but whitespace are ignored.

--container-image=CONTAINER_IMAGE

Full container image name, which should be pulled onto VM instance, eg. docker.io/tomcat.

--container-mount-host-path=[host-path=HOSTPATH,mount-path=MOUNTPATH[,mode=MODE],...]

Mounts a volume by using host-path.

host-path

Path on host to mount from.

mount-path

Path on container to mount to. Mount paths with spaces and commas (and other special characters) are not supported by this command.

mode

Volume mount mode: rw (read/write) or ro (read-only).
Default: rw.

--container-mount-tmpfs=[mount-path=MOUNTPATH,...]

Mounts empty tmpfs into container at MOUNTPATH.

mount-path

Path on container to mount to. Mount paths with spaces and commas (and other special characters) are not supported by this command.

--container-privileged

Specify whether to run container in privileged mode.
Default: --no-container-privileged.

--container-restart-policy=POLICY; default="always"

Specify whether to restart a container on exit. POLICY must be one of: never, on-failure, always.

--container-stdin

Keep container STDIN open even if not attached.
Default: --no-container-stdin.

--container-tty

Allocate a pseudo-TTY for the container.
Default: --no-container-tty.

--create-disk=[PROPERTY=VALUE,...]

Creates and attaches persistent disks to the instances.

name

Specifies the name of the disk. This option cannot be specified if more than one instance is being created.

description

Optional textual description for the disk being created.

mode

Specifies the mode of the disk. Supported options are ro for read-only and rw for read-write. If omitted, rw is used as a default.

image

Specifies the name of the image that the disk will be initialized with. A new disk will be created based on the given image. To view a list of public images and projects, run $ gcloud compute images list. It is best practice to use image when a specific version of an image is needed. If both image and image-family flags are omitted a blank disk will be created.

image-family

The family of the image that the disk will be initialized with. When a family is specified instead of an image, the latest non-deprecated image associated with that family is used. It is best practice to use image-family when the latest version of an image is needed.

image-project

The project that the image or image family belongs to. It is best practice to define image-project.

*

If specifying one of our public images, image-project must be provided.

*

If there are several of the same image-family value in multiple projects, image-project must be specified to clarify the image to be used.

*

If not specified and either image or image-family is provided, the current default project is used.

size

The size of the disk. The value must be a whole number followed by a size unit of KB for kilobyte, MB for megabyte, GB for gigabyte, or TB for terabyte. For example, 10GB will produce a 10 gigabyte disk. Disk size must be a multiple of 1 GB. If not specified, the default image size will be used for the new disk.

type

The type of the disk. To get a list of available disk types, run $ gcloud compute disk-types list. The default disk type is pd-standard.

device-name

An optional name that indicates the disk name the guest operating system will see. If omitted, a device name of the form persistent-disk-N will be used.

auto-delete

If yes, this persistent disk will be automatically deleted when the instance is deleted. However, if the disk is later detached from the instance, this option won't apply. The default value for this is no.

--description=DESCRIPTION

Specifies a textual description of the instances.

--disk=[auto-delete=AUTO-DELETE],[boot=BOOT],[device-name=DEVICE-NAME],[mode=MODE],[name=NAME],[scope=SCOPE]

Attaches persistent disks to the instances. The disks specified must already exist.

name

The disk to attach to the instances. When creating more than one instance and using this property, the only valid mode for attaching the disk is read-only (see mode below).

mode

Specifies the mode of the disk. Supported options are ro for read-only and rw for read-write. If omitted, rw is used as a default. It is an error for mode to be rw when creating more than one instance because read-write disks can only be attached to a single instance.

boot

If yes, indicates that this is a boot disk. The virtual machines will use the first partition of the disk for their root file systems. The default value for this is no.

device-name

An optional name that indicates the disk name the guest operating system will see. If omitted, a device name of the form persistent-disk-N will be used.

auto-delete

If yes, this persistent disk will be automatically deleted when the instance is deleted. However, if the disk is later detached from the instance, this option won't apply. The default value for this is no.

scope

Can be zonal or regional. If zonal, the disk is interpreted as a zonal disk in the same zone as the instance (default). If regional, the disk is interpreted as a regional disk in the same region as the instance. The default value for this is zonal.

--labels=[KEY=VALUE,...]

List of label KEY=VALUE pairs to add.
Keys must start with a lowercase character and contain only hyphens (-), underscores (_), lowercase characters, and numbers. Values must contain only hyphens (-), underscores (_), lowercase characters, and numbers.

--machine-type=MACHINE_TYPE

Specifies the machine type used for the instances. To get a list of available machine types, run 'gcloud compute machine-types list'. If unspecified, the default type is n1-standard-1.

--maintenance-policy=MAINTENANCE_POLICY

Specifies the behavior of the instances when their host machines undergo maintenance. The default is MIGRATE. MAINTENANCE_POLICY must be one of:

MIGRATE

The instances should be migrated to a new host. This will temporarily impact the performance of instances during a migration event.

TERMINATE

The instances should be terminated.

--metadata=KEY=VALUE,[KEY=VALUE,...]

Metadata to be made available to the guest operating system running on the instances. Each metadata entry is a key/value pair separated by an equals sign. Metadata keys must be unique and less than 128 bytes in length. Values must be less than or equal to 32,768 bytes in length. Multiple arguments can be passed to this flag, e.g., --metadata key-1=value-1,key-2=value-2,key-3=value-3.
In images that have Compute Engine tools installed on them, such as the official images (cloud.google.com/compute/docs/images the following metadata keys have special meanings:

startup-script

Specifies a script that will be executed by the instances once they start running. For convenience, --metadata-from-file can be used to pull the value from a file.

startup-script-url

Same as startup-script except that the script contents are pulled from a publicly-accessible location on the web.

--metadata-from-file=KEY=LOCAL_FILE_PATH,[...]

Same as --metadata except that the value for the entry will be read from a local file. This is useful for values that are too large such as startup-script contents.

--min-cpu-platform=PLATFORM

When specified, the VM will be scheduled on host with specified CPU architecture or a newer one. To list available CPU platforms in given zone, run:

$ gcloud compute zones describe ZONE \
  --format="value(availableCpuPlatforms)"

Default setting is "AUTOMATIC".
CPU platform selection is available only in selected zones.
You can find more information on-line: cloud.google.com/compute/docs/instances/specify-min-cpu-platform

--network=NETWORK

Specifies the network that the instances will be part of. If --subnet is also specified subnet must be a subnetwork of network specified by --network. If neither is specified, this defaults to the "default" network.

--network-interface=[PROPERTY=VALUE,...]

Adds a network interface to the instance. Mutually exclusive with any of these flags: --address, --network, --network-tier, --subnet, --private-network-ip.
The following keys are allowed:

address

Assigns the given external address to the instance that is created. Specifying an empty string will assign an ephemeral IP. Mutually exclusive with no-address. If neither key is present the instance will get an ephemeral IP.

network

Specifies the network that the interface will be part of. If subnet is also specified it must be subnetwork of this network. If neither is specified, this defaults to the "default" network.

no-address

If specified the interface will have no external IP. Mutually exclusive with address. If neither key is present the instance will get an ephemeral IP.

network-tier

Specifies the network tier of the interface. NETWORK_TIER must be one of: PREMIUM, STANDARD. The default value is PREMIUM.

private-network-ip

Assigns the given RFC1918 IP address to the interface.

subnet

Specifies the subnet that the interface will be part of. If network key is also specified this must be a subnetwork of the specified network.

aliases

Specifies the IP alias ranges to allocate for this interface. If there are multiple IP alias ranges, they are separated by semicolons.
For example:

--aliases="10.128.1.0/24;range1:/32"

Each IP alias range consists of a range name and an IP range separated by a colon, or just the IP range. The range name is the name of the range within the network interface's subnet from which to allocate an IP alias range. If unspecified, it defaults to the primary IP range of the subnet. The IP range can be a CIDR range (e.g. 192.168.100.0/24), a single IP address (e.g. 192.168.100.1), or a netmask in CIDR format (e.g. /24). If the IP range is specified by CIDR range or single IP address, it must belong to the CIDR range specified by the range name on the subnet. If the IP range is specified by netmask, the IP allocator will pick an available range with the specified netmask and allocate it to this network interface.

--network-tier=NETWORK_TIER

Specifies the network tier that will be used to configure the instance. NETWORK_TIER must be one of: PREMIUM, STANDARD. The default value is PREMIUM.

--preemptible

If provided, instances will be preemptible and time-limited. Instances may be preempted to free up resources for standard VM instances, and will only be able to run for a limited amount of time. Preemptible instances can not be restarted and will not migrate.

--private-network-ip=PRIVATE_NETWORK_IP

Specifies the RFC1918 IP to assign to the instance. The IP should be in the subnet or legacy network IP range.

--restart-on-failure

The instances will be restarted if they are terminated by Compute Engine. This does not affect terminations performed by the user. Enabled by default, use --no-restart-on-failure to disable.

--source-instance-template=SOURCE_INSTANCE_TEMPLATE

The name of the instance template that the instance will be created from.
Users can also override machine type and labels. Values of other flags will be ignored and --source-instance-template will be used instead.

--subnet=SUBNET

Specifies the subnet that the instances will be part of. If --network is also specified subnet must be a subnetwork of network specified by --network.

--tags=TAG,[TAG,...]

Specifies a list of tags to apply to the instance. These tags allow network firewall rules and routes to be applied to specified VM instances. See gcloud compute firewall-rules create(1) for more details.
To read more about configuring network tags, read this guide: cloud.google.com/vpc/docs/add-remove-network-tags
To list instances with their respective status and tags, run:

$ gcloud compute instances list \
    --format='table(name,status,tags.list())'

To list instances tagged with a specific tag, tag1, run:

$ gcloud compute instances list --filter='tags:tag1'

--zone=ZONE

Zone of the instances to create. If not specified, you may be prompted to select a zone. gcloud will attempt to identify the zone by searching for resources in your project. If the zone cannot be determined, you will then be prompted with all Google Cloud Platform zones.
To avoid prompting when this flag is omitted, you can set the compute/zone property:

$ gcloud config set compute/zone ZONE

A list of zones can be fetched by running:

$ gcloud compute zones list

To unset the property, run:

$ gcloud config unset compute/zone

Alternatively, the zone can be stored in the environment variable CLOUDSDK_COMPUTE_ZONE.

At most one of these may be specified:

--address=ADDRESS

Assigns the given external address to the instance that is created. The address may be an IP address or the name or URI of an address resource. This option can only be used when creating a single instance.

--no-address

If provided, the instances will not be assigned external IP addresses.

Custom machine type extensions.

--custom-cpu=CUSTOM_CPU

A whole number value indicating how many cores are desired in the custom machine type. This flag must be specified if any of the other arguments in this group are specified.

--custom-memory=CUSTOM_MEMORY

A whole number value indicating how much memory is desired in the custom machine type. A size unit should be provided (eg. 3072MB or 9GB) - if no units are specified, GB is assumed. This flag must be specified if any of the other arguments in this group are specified.

--custom-extensions

Use the extended custom machine type.

--image-project=IMAGE_PROJECT

The project against which all image and image family references will be resolved. It is best practice to define image-project.

---

If specifying one of our public images, image-project must be provided.

---

If there are several of the same image-family value in multiple projects, image-project must be specified to clarify the image to be used.

---

If not specified and either image or image-family is provided, the current default project is used.

At most one of these may be specified:

--image=IMAGE

Specifies the boot image for the instances. For each instance, a new boot disk will be created from the given image. Each boot disk will have the same name as the instance. To view a list of public images and projects, run $ gcloud compute images list. It is best practice to use --image when a specific version of an image is needed.
When using this option, --boot-disk-device-name and --boot-disk-size can be used to override the boot disk's device name and size, respectively.

--image-family=IMAGE_FAMILY

The family of the image that the boot disk will be initialized with. When a family is specified instead of an image, the latest non-deprecated image associated with that family is used. It is best practice to use --image-family when the latest version of an image is needed.
By default, debian-9 is assumed for this flag.

At most one of these may be specified:

--public-dns

Assigns a public DNS name to the instance.

--no-public-dns

If provided, the instance will not be assigned a public DNS name.

At most one of these may be specified:

--public-ptr

Creates a DNS PTR record for the external IP of the instance.

--no-public-ptr

If provided, no DNS PTR record is created for the external IP of the instance. Mutually exclusive with public-ptr-domain.

At most one of these may be specified:

--public-ptr-domain=PUBLIC_PTR_DOMAIN

Assigns a custom PTR domain for the external IP of the instance. Mutually exclusive with no-public-ptr.

--no-public-ptr-domain

If both this flag and --public-ptr are specified, creates a DNS PTR record for the external IP of the instance with the PTR domain name being the DNS name of the instance.

At most one of these may be specified:

--scopes=[SCOPE,...]

If not provided, the instance will be assigned the default scopes, described below.
SCOPE can be either the full URI of the scope or an alias. default scopes are assigned to all instances. Available aliases are:

Alias	URI
bigquery	www.googleapis.com/auth/bigquery
cloud-platform	www.googleapis.com/auth/cloud-platform
cloud-source-repos	www.googleapis.com/auth/source.full_control
cloud-source-repos-ro	www.googleapis.com/auth/source.read_only
compute-ro	www.googleapis.com/auth/compute.readonly
compute-rw	www.googleapis.com/auth/compute
datastore	www.googleapis.com/auth/datastore
default	www.googleapis.com/auth/devstorage.read_only
	www.googleapis.com/auth/logging.write
	www.googleapis.com/auth/monitoring.write
	www.googleapis.com/auth/pubsub
	www.googleapis.com/auth/service.management.readonly
	www.googleapis.com/auth/servicecontrol
	www.googleapis.com/auth/trace.append
gke-default	www.googleapis.com/auth/devstorage.read_only
	www.googleapis.com/auth/logging.write
	www.googleapis.com/auth/monitoring
	www.googleapis.com/auth/service.management.readonly
	www.googleapis.com/auth/servicecontrol
	www.googleapis.com/auth/trace.append
logging-write	www.googleapis.com/auth/logging.write
monitoring	www.googleapis.com/auth/monitoring
monitoring-write	www.googleapis.com/auth/monitoring.write
pubsub	www.googleapis.com/auth/pubsub
service-control	www.googleapis.com/auth/servicecontrol
service-management	www.googleapis.com/auth/service.management.readonly
sql (deprecated)	www.googleapis.com/auth/sqlservice
sql-admin	www.googleapis.com/auth/sqlservice.admin
storage-full	www.googleapis.com/auth/devstorage.full_control
storage-ro	www.googleapis.com/auth/devstorage.read_only
storage-rw	www.googleapis.com/auth/devstorage.read_write
taskqueue	www.googleapis.com/auth/taskqueue
trace	www.googleapis.com/auth/trace.append
userinfo-email	www.googleapis.com/auth/userinfo.email

DEPRECATION WARNING: www.googleapis.com/auth/sqlservice account scope and sql alias do not provide SQL instance management capabilities and have been deprecated. Please, use www.googleapis.com/auth/sqlservice.admin or sql-admin to manage your Google SQL Service instances.

--no-scopes

Create instance without scopes

At most one of these may be specified:

--service-account=SERVICE_ACCOUNT

A service account is an identity attached to the instance. Its access tokens can be accessed through the instance metadata server and are used to authenticate applications on the instance. The account can be set using an email address corresponding to the required service account. You can explicitly specify the Compute Engine default service account using the 'default' alias.
If not provided, the instance will get project's default service account.

--no-service-account

Create instance without service account

GCLOUD WIDE FLAGS

EXAMPLES

$ gcloud compute instances create-with-container instance-1 \
         --container-image=gcr.io/google-containers/busybox        \
         --container-command='echo "Hello world"'

$ gcloud compute instances create-with-container instance-1 \
         --container-image=gcr.io/google-containers/busybox
  --container-privileged

NOTES

$ gcloud alpha compute instances create-with-container $ gcloud beta compute instances create-with-container