This page describes known issues that you might encounter while using
Cloud Workstations.

## Unable to start workstation due to Compute Engine resource shortages

If Compute Engine is experiencing resource shortages (stockouts) of the
machine type specified in your workstation configuration, you might be unable
to start your workstation. To address this issue, try one of the following
strategies:

1. Change the machine type in your workstation configuration to a machine type that has available capacity in your region.
2. [Reserve capacity](https://docs.cloud.google.com/workstations/docs/reserve-capacity) for your selected machine type. Cloud Workstations automatically consumes reservations, or you can configure the service to consume a specific reservation.
3. Specify additional machine types in your configuration with the [boost mode](https://docs.cloud.google.com/workstations/docs/boost-workstation) feature. This allows your developers to choose a machine type with capacity in your region when starting the workstation.

## Workbench failed to connect to server

This error indicates the editor is unable to establish a WebSocket connection to
the base editor. This is usually due to an intermediary
proxy or network configuration blocking WebSocket traffic, or blocking traffic
to `*.cloudworkstations.dev` domains.

If you aren't sure how to resolve the traffic issue, contact your network
administrator.

## Connecting to a stopped workstation

If you are unable to connect to a stopped workstation, review your workstation
cluster settings. For Cloud Workstations to automatically restart a stopped
workstation when you connect, your workstation cluster must be configured to use
the default Google Cloud console launcher. For more information, see
[Automatically restart stopped workstations](https://docs.cloud.google.com/workstations/docs/automatically-restart-stopped-workstations).

## Disabling the Cloud Workstations API doesn't delete Cloud Workstations resources

Disabling the Cloud Workstations API doesn't delete Cloud Workstations resources.
To avoid being billed, you must delete the underlying Cloud Workstations
resources.

## Environment variables set on the workstation configuration are not set in SSH sessions

Environment variable set at the workstation configuration or workstation level
are not set in the SSH sessions by default, and instead require the use of a
[custom image](https://docs.cloud.google.com/workstations/docs/customize-container-images#container_image_that_sets_container_environment_variables_in_ssh_sessions).

## IntelliJ IDEA-based IDEs

The following features of the IntelliJ IDEA-based IDEs extension are
not available, due to
[limitations on JetBrains Gateway](https://youtrack.jetbrains.com/issue/CWM-838/Editor-like-synchronization-for-consoles):

- Debug on Kubernetes
- Debug a locally running Cloud Run service

## Google Cloud CLI can't find my command

If the `gcloud` CLI can't find your command, update to the current
version:

    gcloud components update

Also, make sure that you use current Cloud Workstations `gcloud` CLI syntax.
For help, see
[`gcloud workstations --help`](https://docs.cloud.google.com/sdk/gcloud/reference/workstations).
As noted in the `gcloud` CLI
[release levels](https://docs.cloud.google.com/sdk/gcloud#release_levels), alpha and beta commands may change
without notice.

## Switching from Regional Persistent Disk to Hyperdisk

It is not possible to switch a workstation configuration from Regional Persistent Disk
to Hyperdisk after the workstation configuration is created.
To switch a workstation's backing home disk, you must create a new configuration using
Hyperdisk and then manually
[clone the workstation](https://docs.cloud.google.com/workstations/docs/clone-workstation) to the new configuration.

## APT repository GPG key expiration or rotation errors

When you run `apt-get update` or install packages inside a workstation
container, you might encounter GPG key errors, such as `KEYEXPIRED` or
`NO_PUBKEY` *KEY_ID*. This occurs when an upstream third-party repository
owner (such as Yarn, NodeSource, or Microsoft) updates or rotates their
signing key out-of-band.

To update the preinstalled APT repository keys in your container, run the key
refresh script:

    sudo /google/scripts/refresh-preinstalled-apt-keys.sh

You can also run this script directly in your custom `Dockerfile` before running `apt-get update` or `apt-get install`. For more information, see [Update preconfigured Secure APT keys](https://docs.cloud.google.com/workstations/docs/customize-container-images#update_preconfigured_secure_apt_keys).