code-server on Cloud Run — Lab Guide

Overview

Estimated time: 45–90 minutes

code-server is Coder's open-source build of Visual Studio Code that runs on a remote server and is accessed entirely through the browser — a full IDE with the extension marketplace, integrated terminal, and a persistent workspace. This lab takes you through the full operational lifecycle of the code-server on Cloud Run module on Google Cloud: deploy it, access and verify it, run it day-to-day, observe it, diagnose common problems, and tear it down.

The lab focuses on operating the Cloud Run module and the Google Cloud platform, not on code-server product features. For the complete list of provisioned services and every configuration input (organised by group), see the Configuration Guide — this lab deliberately does not duplicate that detail so it stays accurate over time.

Objectives

By the end of this lab you will be able to:

Deploy the module from the RAD platform and locate the resources it provisions.
Access the editor through its default public (all) ingress, retrieve the generated password, and verify the service.
Perform day-2 operations — inspect revisions, manage the GCS-backed workspace, and update the version.
Understand why the module is pinned to a single instance and how scaling changes are made safely.
Observe the service with Cloud Logging and Cloud Monitoring.
Diagnose and resolve the most common deployment and runtime issues.
Tear the deployment down cleanly.

Prerequisites

Services_GCP (provides the VPC, Artifact Registry, and shared service accounts this module depends on). You do not need to deploy this yourself first — the platform automatically detects whether it already exists in the target project and provisions it before this module if not (see Task 1).
A Google Cloud project with billing enabled.
gcloud CLI authenticated: gcloud auth login and gcloud auth application-default login.
Project Owner (or equivalent) IAM on the project.
Bringing your own project? Before the first deploy into it, the deployment confirmation dialog asks you to prove you control it (Get verification code, run the commands it shows as a project Owner, then Verify) and to give the RAD deployment service account the Owner role. A project RAD creates for you needs neither.
Advanced mode for later changes. The create form asks only for the first page of inputs (and, in a project RAD creates for you, little more than the tenant name and region). Every other input in the Configuration Guide — including the scaling and version inputs in the Day-2 tasks — is changed afterwards with Update on the deployment's page after ticking Enable advanced mode, which needs a credit balance that covers the update's estimated build cost (updates never carry a module fee). On a lab environment only an administrator can use Advanced mode.
RAD platform access with permission to deploy modules into the project.

Set these shell variables once; every task below reuses them:

export PROJECT="<your-gcp-project-id>"
export REGION="us-central1"          # the region you deploy into

Task 1 — Deploy the module [Automated]

Open Solutions → Solution Catalog → RAD modules in the RAD platform top navigation, open code-server (Cloud Run) from the Platform Modules list to start configuration, choose Configuration Form under How would you like to configure this deployment? (the form opens on the Conversational Assistant if you hold purchased credits or are a partner or administrator), set project_id, and review the inputs. Configure only what you need — the Configuration Guide documents every input by group, with defaults. Click Deploy Module, review the estimated cost in the Deployment Confirmation dialog when it appears and click Submit (if the dialog then adds a confirmation step, such as verifying a project you bring, complete it and click Confirm), which opens the deployment status page with real-time logs.
The platform builds a thin wrapper image over codercom/code-server (mirrored into Artifact Registry via Cloud Build), provisions the Cloud Run service (port 8080, 1 vCPU / 1 GiB, gen2 execution environment), mounts a dedicated GCS workspace bucket via GCS FUSE at /home/coder, and generates a random editor PASSWORD in Secret Manager. There is no Cloud SQL instance and no Redis — code-server has no database. First deploys typically take 10–20 minutes (the container build dominates — there is no database to wait for).

When it completes, discover the resources with name-agnostic filters (so the commands keep working regardless of the deployment suffix):

SERVICE=$(gcloud run services list --project="$PROJECT" --region="$REGION" \
  --filter="metadata.name~codeserver" --format="value(metadata.name)" --limit=1)
SERVICE_URL=$(gcloud run services describe "$SERVICE" \
  --project="$PROJECT" --region="$REGION" --format="value(status.url)")
echo "Service: $SERVICE"
echo "URL:     $SERVICE_URL"

Task 2 — Access & verify [Manual]

Mind the ingress mode first. The module defaults to ingress_settings = "all" — the editor is publicly reachable out of the box, gated by the auto-generated PASSWORD secret (see step 3). Check the current mode:
```
gcloud run services describe "$SERVICE" --project="$PROJECT" --region="$REGION" \
  --format="value(metadata.annotations['run.googleapis.com/ingress'])"
```
To restrict access to the VPC instead, set ingress_settings = "internal" via Update on the deployment details page — a curl from outside the VPC will then return 404 (the ingress policy working, not a failure). Whenever ingress is all, keep enable_password = true (a public, unauthenticated IDE includes a public terminal).
Once reachable, confirm the service is healthy. code-server's unauthenticated health path is /healthz (note: not /health, which returns 401 when a password is set):
```
curl -s -o /dev/null -w "%{http_code}\n" "$SERVICE_URL/healthz"
```

Retrieve the generated editor password from Secret Manager, then open $SERVICE_URL in a browser and log in:

PW_SECRET=$(gcloud secrets list --project="$PROJECT" \
  --filter="name~codeserver AND name~password" --format="value(name)" --limit=1)
gcloud secrets versions access latest --secret="$PW_SECRET" --project="$PROJECT"

Verify persistence: create a file or install an extension in the editor, then confirm it lands in the workspace bucket — everything under /home/coder (settings, keybindings, extensions, open projects) lives on GCS FUSE:
```
WORKSPACE_BUCKET=$(gcloud storage buckets list --project="$PROJECT" \
  --filter="name~codeserver" --format="value(name)" --limit=1)
gcloud storage ls "gs://$WORKSPACE_BUCKET/"
```

Task 3 — Operate & keep it running (Day-2) [Manual]

Inspect the service and its revisions (each deploy creates an immutable revision; traffic shifts to the newest healthy one):

gcloud run services describe "$SERVICE" --project="$PROJECT" --region="$REGION"
gcloud run revisions list --service="$SERVICE" --project="$PROJECT" --region="$REGION"

Do not scale out. The module deliberately pins min_instance_count = max_instance_count = 1: editor sessions are held in memory and the workspace volume has a single writer — a second instance would split sessions and risk concurrent writes to /home/coder. Resource changes (cpu_limit, memory_limit for heavy language servers) go through Update on the deployment details page, not manual gcloud edits (a manual edit would be reverted on the next apply).
Update the application version by changing the version input via Update on the deployment details page; a new image builds and a new revision rolls out. There are no migrations — code-server has no schema. latest pins to 4.99.1 at build time; pin a specific release in production.

Manage secrets and storage:

gcloud secrets list --project="$PROJECT" --filter="name~codeserver"
gcloud storage buckets list --project="$PROJECT" --filter="name~codeserver"
# One-off workspace backup:
gcloud storage cp -r "gs://$WORKSPACE_BUCKET" "gs://<your-backup-bucket>/codeserver-$(date +%F)"

There is no database session to open. database_type = "NONE" — no Cloud SQL instance, no db-init job, no database password. The only durable state is the workspace bucket.

Task 4 — Observe: Logging & Monitoring [Manual]

Logs — from the CLI or the Logs Explorer:
```
gcloud run services logs read "$SERVICE" --project="$PROJECT" --region="$REGION" --limit=50
```
Logs Explorer filter: resource.type="cloud_run_revision" AND resource.labels.service_name="<service>".
Monitoring — open the Cloud Run dashboard for the service and review request count, request latency (P50/P95/P99), instance count (should sit flat at 1), and CPU / memory utilisation — language servers and extensions are the usual memory drivers. The module's uptime check is disabled by default (uptime_check_config.enabled = false) and additionally requires a publicly reachable endpoint, so Monitoring → Uptime checks may legitimately be empty even though ingress defaults to all.

Task 5 — Troubleshoot & debug [Manual]

Durable techniques for the failure modes you are most likely to hit. These are platform-level diagnostics and do not change with code-server releases.

URL returns 404 from your machine: if ingress_settings was changed to internal, this is expected — not an outage. Check the ingress annotation (Task 2) before reading logs.
Revision never becomes Ready: check the probe paths. Probes must target the unauthenticated /healthz; pointing them at /health while a password is set returns 401 and the revision fails readiness even though the app booted fine:
```
gcloud run revisions list --service="$SERVICE" --project="$PROJECT" --region="$REGION"
gcloud run services logs read "$SERVICE" --project="$PROJECT" --region="$REGION" --limit=100
```
Login rejected: re-read the PASSWORD secret (Task 2, step 3) — the value is injected as the container's PASSWORD env var; a new secret version only takes effect on the next revision.
Workspace state missing / extensions gone: confirm the workspace bucket exists and the execution environment is gen2 (GCS FUSE cannot mount under gen1 — the plan-time validation catches this, but verify if the module was overridden).
Editor slow / OOM kills: raise memory_limit (heavy language servers can OOM below 1 GiB) via the RAD Update flow and watch the memory chart in Monitoring.
Image build failed: review Cloud Build history for the failed build's log; the image is a thin wrapper over codercom/code-server mirrored into Artifact Registry.
403 / permission errors: verify the runtime service account's IAM roles.

See the Configuration Guide's Configuration Pitfalls section for setting-specific gotchas.

Task 6 — Tear down [Automated]

On the Deployments page, open the deployment and click the Trash icon (Delete). Delete runs terraform destroy and is irreversible (the deployment record is retained for history). If a deployment is stuck and the RAD platform can no longer manage it (for example after manual changes that conflict with the Terraform state), use Purge instead (from the same Delete dialog) — it removes the deployment from RAD's records without destroying the cloud resources (it makes RAD forget the deployment). This removes everything the module created — the Cloud Run service, the Secret Manager PASSWORD secret, the GCS workspace bucket (and with it all files, settings, and extensions under /home/coder), and Artifact Registry images. Copy the workspace bucket first if you want to keep your work. Resources owned by Services_GCP (the VPC, shared Cloud SQL, registry) are managed separately and are not removed here.

Summary

Task	Type	Outcome
1 — Deploy	Automated	Module builds the image and provisions Cloud Run + the GCS workspace bucket + the `PASSWORD` secret (no DB, no Redis)
2 — Access & verify	Manual	Understand the default public (`all`) ingress; `/healthz` passes; retrieve the password and log in; verify workspace persistence
3 — Operate	Manual	Inspect revisions, keep single-instance scaling, update version, back up the workspace bucket
4 — Observe	Manual	Query Cloud Logging; review Cloud Monitoring metrics; understand when the uptime check exists
5 — Troubleshoot	Manual	Diagnose ingress, probe-path, password, workspace, memory, build, and IAM issues
6 — Tear down	Automated	Delete (Trash) removes all module resources including the workspace bucket

Need RAD to do something it does not do yet? Request it on the roadmap, or vote on what is already there.

By Dr Shiyghan Emmanuel Navti · Google Cloud certified: ACE · PCA · PCD · PDE · PSE · CDL

Overview​

Objectives​

Prerequisites​

Task 1 — Deploy the module [Automated]​

Task 2 — Access & verify [Manual]​

Task 3 — Operate & keep it running (Day-2) [Manual]​

Task 4 — Observe: Logging & Monitoring [Manual]​

Task 5 — Troubleshoot & debug [Manual]​

Task 6 — Tear down [Automated]​

Summary​