Budibase on Google Cloud Run
Budibase is an open-source low-code platform for building internal tools, business apps, and workflows on top of your data. This module deploys Budibase on Cloud Run v2 on top of the App_CloudRun foundation, which provisions and manages the shared Google Cloud infrastructure.
This guide focuses on the cloud services Budibase uses and how to explore and operate them from the Google Cloud Console and the command line. For the mechanics common to every Cloud Run application — service identity, ingress and load balancing, scaling and concurrency, CI/CD, Cloud Armor, IAP, Binary Authorization, VPC Service Controls, backups, and the deployment lifecycle — refer to the App_CloudRun foundation guide rather than repeating them here.
What Budibase costs on RAD, and how that compares
Budibase on RAD's Cloud Run module costs about US$232 a month in your own Google Cloud project, with no extra licence fee — Budibase is open source. In a project RAD manages for you, the same deployment is metered at 105 credits a day (about 3,150 a month, about US$315 at the top-up price). Or see the GKE guide for always-on capacity across pods. If you only need Budibase occasionally — studying, a demo, a short-lived project — a RAD-managed project can be deleted and restored within 30 days for a few credits, so that monthly figure becomes a few dollars instead. See Pause it for free, below.
What you pay on RAD
Budibase is open source, so there is no licence fee. You pay for three things:
| In a Google Cloud project you own | In a project RAD manages for you | |
|---|---|---|
| Module fee, once per deployment | 300 credits (US$30 at the top-up price) | 270 credits (10% lower) |
| Build time | A few credits per build (roughly 3–6 credits) | The same |
| Google Cloud running cost | Billed by Google to your own billing account, about US$232 a month for the default configuration (table below) | Metered hourly in credits; RAD publishes 105 credits a day, about 3,150 a month (about US$315 at the top-up price) |
| Who owns what | You own the project and its billing; RAD deploys and updates it | RAD owns the project, with guardrails, quotas and budget alerts; you get console access, and a minimum purchased balance is held in reserve |
10 credits cost US$1 on a one-off top-up, and less on a monthly plan. In a project RAD manages, the database, file server and network are shared by every application in the project, so a second application does not add a second database.
Default running cost in your own project (us-central1, Google list prices):
| Resource (module default) | Per month |
|---|---|
| Cloud Run, 4x vCPU / 8 GiB (alwaysOn) | US$226.05 |
| Cloud NAT and networking | US$5 |
| Cloud Storage (add-ons, backups) | US$1 |
| Total | about US$232 |
These prices are the same in africa-south1 (Johannesburg).
- This module runs continuously by design — there is no scale-to-zero option here — so the Cloud Run figure above already assumes it is always on.
How it compares
Budibase's own team sells a hosted Budibase Cloud:
| Plan | Per month | Notes |
|---|---|---|
| Pro | US$19 | Entry cloud plan |
| Premium | US$49 | More creators, AI credits, log retention |
| Business | US$299 | Higher limits, more workspaces |
| Enterprise | Custom | Contact sales |
Budibase Cloud bills by plan tier (creators, AI credits, workspaces), not by the compute this app actually uses, so it is not a like-for-like comparison with the table above — it is the official alternative if you would rather not run any infrastructure at all. RAD's module runs the same open-source Budibase on infrastructure you can see and that scales with your own usage rather than a plan's seat/creator limits.
Pause it for free: delete a RAD-managed project, restore it when you need it
If Budibase runs in a project RAD manages for you, deleting the whole project — rather than deleting just this module — unlinks its billing first, then asks Google to delete the project outright. Google keeps it, recoverable, for 30 days, and because billing is already unlinked, nothing is charged while it waits. Unlike deleting one module, this does not tear down Cloud SQL, any VM or the compute resource one by one: the whole project simply stops.
- Restoring is a check, not a rebuild. Within 30 days, the project's owner can ask Google to undelete the project and reattach its billing account, then run Update on each deployment to confirm everything came back. Because nothing was individually destroyed, that Update finds the same resources already there, and an Update never charges the module fee again. For a typical 2–3-deployment chain this costs only a handful of credits in total (roughly 5–10 credits, under US$1).
- What this needs. You must own the project (not one RAD only manages billing for), restore it yourself within 30 days — after that Google deletes it for good — and clear the tier's admission floor again (100 credits for the sandbox tier most study/demo use fits). Google says most services are fully working again within 36 hours of a restore.
- One real gap: nightly backups do not survive. Backups are written to a bucket inside the project, and that bucket has Cloud Storage's soft-delete turned off, so it is very likely gone as soon as the project is deleted — even though the project itself is recoverable for 30 days. Copy a backup out (to Google Drive, or a bucket outside the project) before deleting if there is anything in it worth keeping; for a default install with nothing irreplaceable, this does not matter.
Source: Delete and restore projects.
Pay only while you use it, the other way: delete and redeploy
In your own project, or once the 30-day restore window above has passed, the way to stop paying is to delete the deployment and deploy it again when you need it.
- What a redeploy costs. The module fee again, plus the builds: about 304 credits (about US$30.40) at the top-up price, and roughly 15–20 minutes of build time (RAD's own platform average). Deleting saves money only once Budibase would otherwise sit unused for about 3 days or more — in your own project that is about US$7.74 a day, and in a RAD-managed one it is 105 credits a day.
- This module's running cost is entirely its own — nothing in the table above is shared with other applications in the project, so deleting Budibase stops all of it.
- Keep data first. Nightly backups go to a bucket inside the deployment and are deleted with it, so copy the latest backup out before deleting if you want to keep it.
Lab sessions and Managed Environments
- Lab sessions, for training. A trainer runs a session for a class. Each participant gets the app in their own Google Cloud project for 15 minutes to 24 hours, within an allowance the trainer sets. Either the trainer funds every place, or each participant pays for their own. Everything is deleted when the session ends and unused credits go back to the trainer.
- Managed Environments, for consultancies. A partner runs the app for a client from a ring-fenced wallet it funds, and settles with the client directly. If the wallet runs low, billing pauses and the data is kept, so nobody receives an unexpected charge. At the end the partner hands the project over and the deployments become the client's own.
Credits can be bought in more than 20 currencies, including XAF, XOF, NGN, GHS, KES and ZAR, by card, bank transfer or mobile money.
Sources (8 October 2026): Google Cloud list prices from the Cloud Billing Catalog API; RAD fees and the daily-credit estimate from radmodules.dev/pricing; Delete and restore projects for the pause/restore window and backup-bucket caveat; Budibase Cloud pricing; prices change, check each source before relying on a figure.
1. Overview
Budibase runs as a single all-in-one container on Cloud Run v2. The official
budibase/budibase image bundles CouchDB + MinIO + Redis and the Budibase
apps/worker/proxy together and serves HTTP on port 80 — there is no external
managed database. The deployment wires together a focused set of Google Cloud
services:
| Capability | Google Cloud service | Notes |
|---|---|---|
| Compute | Cloud Run v2 | Single all-in-one container, 4 vCPU / 8 GiB by default; runs as one instance (min = max = 1) |
| Database | None (bundled CouchDB) | database_type = "NONE" — CouchDB, MinIO, and Redis all run inside the container |
| Object storage | Cloud Storage | One data bucket provisioned automatically; Budibase's own asset store is the bundled MinIO |
| Cache & queue | Bundled Redis | Runs inside the container on loopback; enable_redis is off by default |
| Secrets | Secret Manager | Seven auto-generated internal credentials injected as service secret env vars |
| Ingress | Cloud Run URL / Cloud Load Balancing | Default run.app URL; optional external HTTPS load balancer + custom domain |
Sensible defaults worth knowing up front:
- Cloud Run is ephemeral / demo-only for Budibase. All state (CouchDB documents
- MinIO objects) lives on the container path
/data, and Cloud Run has no durable local disk — a restart or new revision loses the data store. For a persistent deployment use the GKE variant, which mounts a block PVC at/data.
- MinIO objects) lives on the container path
- Runs as a single instance.
min_instance_count = 1andmax_instance_count = 1. The all-in-one container holds all state locally, so multiple replicas would not share data (split-brain). Scale-to-zero would drop the running data store, so min is 1. cpu_always_allocated = true. CPU is allocated at all times so the bundled CouchDB/MinIO/Redis background processes keep running between requests; keep ittruewithmin_instance_count >= 1.- Seven internal credentials are generated automatically and stored in Secret
Manager (
INTERNAL_API_KEY,JWT_SECRET,MINIO_ACCESS_KEY,MINIO_SECRET_KEY,API_ENCRYPTION_KEY,REDIS_PASSWORD,COUCH_DB_PASSWORD). These must never be rotated after first boot — the data on/datais keyed with them and becomes unreadable if they change. - Port 80 is fixed. The all-in-one image's nginx proxy serves the whole app on
port 80, so
container_portand the probes are pinned to 80. - No external database or
db-initjob. Budibase self-provisions CouchDB and MinIO on first boot;enable_cloudsql_volumeanddatabase_typedefault to off/NONE. - Public ingress by default.
ingress_settings = "all"so the app is reachable at itsrun.appURL; enabling IAP puts Google sign-in in front of it.
2. Google Cloud Services & How to Explore Them
All commands assume PROJECT and REGION are set. Service and resource names are
reported in the deployment Outputs.
A. Cloud Run — the Budibase service
Budibase runs as a Cloud Run v2 service. Because it holds all state locally it runs as a single instance (min = max = 1); each deployment creates an immutable revision.
- Console: Cloud Run → select the service for revisions, traffic, logs, and metrics.
- CLI:
gcloud run services list --project "$PROJECT" --region "$REGION"
gcloud run services describe <service-name> --project "$PROJECT" --region "$REGION"
gcloud run revisions list --service <service-name> --project "$PROJECT" --region "$REGION"
See App_CloudRun for scaling, concurrency, execution environment, and traffic splitting.
B. Data store (bundled CouchDB + MinIO)
There is no Cloud SQL instance — database_type = "NONE". Budibase's CouchDB
document store and MinIO object store both run inside the container and persist
to /data. On Cloud Run that directory is on the container's ephemeral filesystem,
so it does not survive a restart. Inspect the bundled services through the container
env and logs rather than a managed DB console:
- CLI:
# Confirm database_type=NONE and the bundled-service env in the running revision:
gcloud run services describe <service-name> --region "$REGION" \
--format='value(spec.template.spec.containers[0].env)'
For a durable data store, use the GKE variant (block PVC on /data).
C. Cloud Storage
A dedicated Cloud Storage bucket (name suffix storage) is provisioned
automatically. Budibase's own asset/attachment store is the bundled MinIO; this GCS
bucket is available for foundation-level storage integration.
- Console: Cloud Storage → Buckets.
- CLI:
gcloud storage buckets list --project "$PROJECT"
gcloud storage ls gs://<data-bucket>/ # bucket name is in the Outputs
See App_CloudRun for GCS Fuse and CMEK options.
D. Redis (bundled)
Redis runs inside the all-in-one container on loopback and is authenticated with
the auto-generated REDIS_PASSWORD. enable_redis is off by default — do not
enable an external Redis unless you are deliberately externalising the cache.
- CLI:
# Verify the bundled Redis password secret is wired into the revision:
gcloud run services describe <service-name> --region "$REGION" \
--format='value(spec.template.spec.containers[0].env)' | tr ',' '\n' | grep -i redis
E. Secret Manager
Seven internal credentials are generated automatically and stored in Secret Manager,
then injected as service secret env vars: INTERNAL_API_KEY, JWT_SECRET,
MINIO_ACCESS_KEY, MINIO_SECRET_KEY, API_ENCRYPTION_KEY, REDIS_PASSWORD, and
COUCH_DB_PASSWORD. They must never be rotated after first boot.
- Console: Security → Secret Manager.
- CLI:
gcloud secrets list --project "$PROJECT" --filter="name~budibase"
gcloud secrets versions access latest --secret=<secret-name> --project "$PROJECT"
See App_CloudRun for injection details and Budibase_Common for what each secret protects.
F. Networking & ingress
The service is reachable at its run.app URL by default. An external HTTPS load
balancer with a custom domain, Cloud CDN, and Cloud Armor can be layered on; ingress
settings and VPC egress control connectivity.
- Console: Cloud Run (service URL); Network services → Load balancing.
- CLI:
gcloud run services describe <service-name> --region "$REGION" --format='value(status.url)'
gcloud compute addresses list --project "$PROJECT"
See App_CloudRun.
G. Cloud Logging & Monitoring
Container logs flow to Cloud Logging; Cloud Run metrics flow to Cloud Monitoring, with optional uptime checks and alert policies.
- Console: Logging → Logs Explorer; Monitoring → Dashboards / Alerting.
- CLI:
gcloud run services logs read <service-name> --project "$PROJECT" --region "$REGION" --limit 50
3. Budibase Application Behaviour
- No external database bootstrap. With
database_type = "NONE"there is nodb-initjob. Budibase self-provisions its bundled CouchDB and MinIO on first boot inside the container. Only user-suppliedinitialization_jobsare honoured. - State lives on
/data— and Cloud Run does not persist it. CouchDB documents and MinIO objects are written to/dataon the container's ephemeral filesystem. Any restart or new revision starts from an empty store, so treat a Cloud Run Budibase deployment as ephemeral/demo-only and use the GKE variant for anything you need to keep. - Internal credentials are immutable after first boot. The seven generated
secrets key the data on
/data. ChangingAPI_ENCRYPTION_KEYcorrupts all encrypted stored data; changingJWT_SECRETinvalidates all sessions; changing the MinIO or CouchDB credentials breaks access to the object/document stores. Only rotate during a planned reset. - First-run setup. Budibase self-hosted ships with no default admin account. Open the service URL after deploy and create the initial administrator (email + password) through the setup screen before use.
- Health path. The startup probe is TCP on port 80 (not HTTP): nginx binds
the port within seconds but returns 502 until the bundled CouchDB/MinIO/app
upstreams finish booting, so an HTTP check on
/would fail for the whole upstream-boot window — TCP passes as soon as nginx binds the port, letting the revision go Ready while the upstreams finish in the background (30-second initial delay, 40-retry window at a 15-second period, ~10 minutes total). The liveness and readiness probes are HTTP on the unauthenticated root/; liveness uses a 240-second initial delay to clear the post-startup window where/still 502s. - Single-instance only. Keep
min_instance_count = max_instance_count = 1; the data store is local to the container and cannot be shared across replicas. - Verify the running revision:
gcloud run services describe <service-name> --region "$REGION" --project "$PROJECT" \
--format='value(status.url)'
gcloud run jobs list --project "$PROJECT" --region "$REGION" # only user-supplied init jobs
4. Configuration Variables
Variables are grouped exactly as they appear on the deployment platform. Only settings specific to or notable for Budibase are listed; every other input is inherited from App_CloudRun with its standard behaviour.
Group 3 — Application Identity
| Variable | Default | Description |
|---|---|---|
application_name | budibase | Base name for resources. Do not change after first deploy. |
display_name | Budibase | Human-readable name shown in the Console. |
application_version | 3.39.29 | Budibase image tag; used as FROM budibase/budibase:<tag> for the thin wrapper build. Increment to trigger a new build/revision. |
All other inputs follow standard App_CloudRun behaviour.
Group 4 — Runtime & Scaling
| Variable | Default | Description |
|---|---|---|
deploy_application | true | Set false to provision infrastructure only. |
container_port | 80 | The all-in-one image's nginx proxy serves the whole app on port 80 — must stay 80. |
container_resources | 4000m / 8Gi | CPU and memory per instance; the bundled CouchDB/MinIO/Redis + app tier need generous memory — 4Gi OOM-loops on Cloud Run gen2 (the writable /data dir counts against the memory limit), so 8Gi is the reliable minimum. |
min_instance_count | 1 | Keep at 1 — scale-to-zero would drop the local data store. |
max_instance_count | 1 | Keep at 1 — the container holds all state locally; replicas would not share data. |
cpu_always_allocated | true | Allocate CPU at all times so bundled background services keep running between requests. |
execution_environment | gen2 | Gen2 recommended. |
enable_cloudsql_volume | false | Budibase uses no external SQL database. |
enable_image_mirroring | true | Mirror the Budibase image into Artifact Registry. |
All other inputs follow standard App_CloudRun behaviour.
Group 5 — Access & Ingress Control
| Variable | Default | Description |
|---|---|---|
ingress_settings | all | Reachable at the run.app URL by default. |
enable_iap | false | Require Google sign-in in front of Budibase. |
All other inputs follow standard App_CloudRun behaviour.
Group 6 — Environment Variables & Secrets
| Variable | Default | Description |
|---|---|---|
environment_variables | {} | Extra non-secret settings merged into the container. Core values (BUDIBASE_ENVIRONMENT, SELF_HOSTED, COUCH_DB_USER, LOG_LEVEL) are set automatically. |
secret_environment_variables | {} | Map of env var → Secret Manager secret name. The seven internal credentials are injected automatically — do not set them here. |
All other inputs follow standard App_CloudRun behaviour.
Group 12 — Database Backend
| Variable | Default | Description |
|---|---|---|
database_type | NONE | Budibase bundles its own CouchDB; no external managed database is provisioned. |
All other inputs follow standard App_CloudRun behaviour.
Group 21 — Redis Cache & Queue
| Variable | Default | Description |
|---|---|---|
enable_redis | false | Redis runs inside the container; leave off unless externalising the cache. |
redis_host / redis_port / redis_auth | "" / 6379 / "" | Only used if an external Redis is enabled. |
All other inputs follow standard App_CloudRun behaviour.
5. Outputs
Returned on a successful deployment — the quickest way to locate and explore the running resources.
| Output | Description |
|---|---|
service_name | Cloud Run service name. |
service_url | Default run.app URL of the service. |
service_location | Region the service runs in. |
stage_services | Stage-specific service URLs (Cloud Deploy). |
load_balancer_ip / load_balancer_url | External HTTPS load balancer IP / URL (when enabled). |
database_instance_name / database_name / database_user | Populated only if a managed DB is used; empty for Budibase (database_type = NONE). |
database_password_secret / database_host / database_port | DB secret / endpoint / port (unused for Budibase). |
storage_buckets | Created Cloud Storage buckets. |
network_name / network_exists / regions | VPC network, presence, regions. |
container_image / container_registry | Deployed image and Artifact Registry repo. |
monitoring_enabled / monitoring_notification_channels / uptime_check_names | Monitoring status, channels, uptime checks. |
initialization_jobs | Names of any user-supplied init jobs. |
deployment_id / tenant_id / resource_prefix | Naming identifiers. |
project_id / project_number | Project identifiers. |
cicd_enabled / github_repository_url / github_repository_owner / github_repository_name / cicd_configuration | CI/CD status and details. |
artifact_registry_repository / cloudbuild_trigger_name / cloudbuild_trigger_id | Registry and build trigger. |
vpc_sc_enabled / vpc_sc_perimeter_name / vpc_sc_dry_run_mode | VPC-SC status. |
audit_logging_enabled / artifact_registry_cmek_enabled | Audit logging and CMEK status. |
6. Configuration Pitfalls & Sensible Defaults
Risk: Critical (data loss / outage / security) — High (service degraded) — Medium (cost or partial degradation) — Low (minor).
Inherited plan-time validation. This module passes its configuration through the App_CloudRun foundation engine, which validates values and combinations at plan time — an out-of-range
container_port, an invalid CPU/memory quantity, adatabase_typethat does not match an enabled extension, an out-of-rangeredis_port/backup_retention_days. Invalid configuration fails the plan with a clear, named error before any resource is created, so most mistakes below are caught up front rather than at apply or runtime.
| Setting | Sensible value | Risk | Consequence if wrong |
|---|---|---|---|
| Cloud Run as the platform | GKE for persistence | Critical | Cloud Run has no durable disk — /data (all CouchDB + MinIO state) is lost on any restart/revision. Use Budibase_CloudRun only for demo/evaluation. |
API_ENCRYPTION_KEY (auto-generated) | Never rotate after first boot | Critical | Rotating it corrupts all encrypted stored data — it cannot be decrypted. |
MINIO_ACCESS_KEY / MINIO_SECRET_KEY / COUCH_DB_PASSWORD (auto-generated) | Never rotate after first boot | Critical | Rotating breaks access to the bundled object/document stores keyed on /data. |
JWT_SECRET (auto-generated) | Only rotate in a maintenance window | High | Rotating it invalidates all active user sessions, forcing immediate re-login. |
max_instance_count | 1 | Critical | More than one instance splits the local data store — replicas do not share /data (split-brain, data loss). |
min_instance_count | 1 | High | Scale-to-zero drops the running in-container data store. |
cpu_always_allocated | true | High | Request-based billing throttles the bundled CouchDB/MinIO/Redis background work to ~0 CPU between requests. |
container_port | 80 | High | The nginx proxy serves the app on 80; any other port fails the probes and the service never becomes Ready. |
database_type | NONE | High | Selecting an external engine provisions an unused Cloud SQL instance; Budibase never connects to it. |
memory_limit | 8Gi | High | 4Gi OOM-loops (instance restarts ~every 60s) — the writable /data dir (CouchDB/MinIO/Redis state) is in-memory on Cloud Run gen2 and counts against the limit; 8Gi is the reliable minimum. |
| First admin account | Create immediately after deploy | High | Budibase self-hosted ships with no default admin — an unclaimed instance can be claimed by anyone who reaches the URL. |
enable_iap | enable for private deployments | Medium | Without IAP or a WAF the UI is publicly reachable at the run.app URL. |
For the foundation behaviour referenced throughout — service identity, scaling and concurrency, ingress and load balancing, CI/CD, Cloud Armor, IAP, Binary Authorization, VPC-SC, backups, and image mirroring — see App_CloudRun. Budibase-specific application configuration shared with the GKE variant is described in Budibase_Common.
Related guides
- Hands-on lab: Budibase on Cloud Run — deploy it step by step, with the console screens and commands at each stage.
- Budibase on GKE Autopilot — the same application on Kubernetes, for when you need the other deployment target.
- Budibase Common — Shared Application Configuration — the configuration shared by both deployment targets.
Need RAD to do something it does not do yet? Request it on the roadmap, or vote on what is already there.