Filebrowser on GKE Autopilot
File Browser is a lightweight, open-source web file manager written in Go — it serves a directory tree over HTTP for browsing, uploading, editing, and sharing files. This module deploys Filebrowser on GKE Autopilot on top of the App_GKE foundation, which provisions and manages the shared Google Cloud and Kubernetes infrastructure.
This guide focuses on the cloud services Filebrowser uses and how to explore and operate them from the Google Cloud Console and the command line. For the mechanics that are common to every GKE application — Workload Identity, ingress, autoscaling, CI/CD, Cloud Armor, IAP, Binary Authorization, VPC Service Controls, backups, and the deployment lifecycle — refer to the App_GKE foundation guide rather than repeating them here.
What Filebrowser costs on RAD, and how that compares
In a Google Cloud project you own, Filebrowser on RAD's GKE Autopilot module costs about US$42 a month, plus US$7.50 once per deployment and no per-user licence. Filebrowser is a lightweight, self-hosted web file manager with an embedded SQLite database. Figures are as at 8 October 2026; sources are listed at the end of this section.
If this runs in a project RAD manages for you, it can be deleted and restored within 30 days for a few credits instead of sitting there running — so an occasional-use deployment costs a few dollars a month, not US$42. See Pause it for free, below.
What you pay on RAD
| In a Google Cloud project you own | In a project RAD manages for you | |
|---|---|---|
| Module fee, once per deployment | 75 credits (US$7.50 at the top-up price) | 67.5 credits (10% lower) |
| Build time | A few credits per build | The same |
| Google Cloud running cost | Billed by Google to your own billing account, about US$42 a month for the default configuration (table below) | Metered hourly in credits; RAD publishes 67 credits a day, about 2,010 a month (about US$201 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 |
|---|---|
| GKE Autopilot pod, 1 vCPU / 1 GiB, running all the time | US$36 |
| Cloud NAT gateway and IP address | US$5 |
| Cloud Storage for backups | US$1 |
| Total | about US$42 |
GKE's free tier covers the management fee of one Autopilot or zonal cluster per billing account, so the total above assumes this is that cluster; otherwise add about US$73 a month.
How it compares
We could not find a commercial managed/SaaS plan specifically for Filebrowser itself as at 8 October 2026, so the honest comparison is against running it yourself:
| Option | Per month | Who runs it |
|---|---|---|
| Hetzner CPX22 (2 vCPU/4GB) | ~US$24 | You: OS, updates, backups, security patches |
| DigitalOcean 2 vCPU/4GB Droplet | US$24 | You: OS, updates, backups, security patches |
| GCP Compute Engine e2-standard-2 (2 vCPU/8GB) | ~US$49 | You: OS, updates, backups, security patches |
| RAD, GKE Autopilot, your own project | US$42, plus US$7.50 once | RAD's automation, in your project |
Filebrowser's own footprint here (1 vCPU / 1 GiB) is lighter than that reference box, so a smaller, cheaper tier from the same providers would comfortably run it too.
- RAD is usually not cheaper in cash than a bare VPS — the difference is what is managed for you. A VPS price does not include the admin's own time for backups, security patches and OS upgrades; RAD's managed Cloud SQL (where this app uses a database), Secret Manager and monitoring cover that instead.
- Filebrowser is a lightweight, self-hosted web file manager with an embedded SQLite database. That shapes which column matters: predictable, hands-off operation is worth more to most teams running this kind of tool than shaving a few dollars off the monthly bill.
Pause it for free: delete a RAD-managed project, restore it when you need it
If Filebrowser runs in a project RAD manages for you, you have an option beyond scaling down: delete the whole project, and restore it within 30 days for close to nothing. This suits a deployment you only need occasionally — studying for a certification, a demo, a seasonal need — far better than running it continuously.
- How it works. Deleting a RAD-managed project unlinks its billing first, then asks Google to delete the project. Google does not remove it immediately: it 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 GKE Autopilot one by one — the whole project, and everything in it, simply stops.
- Restoring costs a handful of credits, not a rebuild. Within 30 days, the project's owner can restore it. RAD asks Google to undelete the project and reattaches its billing account, then asks you to run Update on each deployment to confirm everything came back. Because nothing was individually destroyed, that Update finds the same resources already there — it is a check, not a rebuild, and an Update never charges the module fee again. That costs a handful of build-time credits in total (under US$1) for a typical 2–3-deployment chain.
- So a month of occasional use can cost a few dollars, not US$42. Deploy Filebrowser, use it for a session, delete the project. Restore it next time you want it, confirm with Update, and delete it again when you're done. You pay only for the module fee once, the builds, and whatever hours it was actually live.
- What this needs. You must own the project (not a bring-your-own one RAD only manages billing for), and you restore it yourself within the 30 days — after that, Google deletes it for good. Restoring is admitted like creating a new project: your purchased credit balance must still clear the tier's floor (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.
Pay only while you use it, the other way: delete and redeploy
GKE keeps at least one pod running, so Filebrowser never scales to zero on its own. The option above only applies to a RAD-managed project; in your own project, or once the 30-day window 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 80 credits (US$8). Deleting saves money if Filebrowser would otherwise sit unused for about 6 days or more in your own project (about US$1.40 a day), or about 2 days or more in a RAD-managed one (67 credits a day).
- Note your settings. RAD does not recreate a deleted deployment for you; you enter the settings again when you deploy.
Lab sessions and Managed Environments
- Lab sessions, for training. A trainer runs a session for a class. Each participant gets Filebrowser 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's own fees and daily-credit estimate from radmodules.dev/pricing; Delete and restore projects for the pause/restore mechanism. Prices change; check each source before relying on a figure.
1. Overview
Filebrowser runs as a single Go web workload. It is deliberately minimal — no SQL database, no cache, no queue — so the deployment wires together a small set of Google Cloud services:
| Capability | Google Cloud service | Notes |
|---|---|---|
| Compute | GKE Autopilot | Single Go pod, 1 vCPU / 1 GiB by default; min = max = 1 |
| Persistent state | Cloud Storage (GCS FUSE) or a block PVC | Mounted at /database; holds the embedded SQLite DB |
| Database | None (embedded SQLite) | database_type = NONE; no Cloud SQL is provisioned |
| Cache & queue | None | Filebrowser uses no Redis |
| Secrets | Secret Manager | No app secrets generated; users live in the SQLite DB |
| Ingress | Cloud Load Balancing | ClusterIP Service by default; custom domain + managed certificate available |
Sensible defaults worth knowing up front:
- State lives in an embedded SQLite file at
/database. Filebrowser has no Cloud SQL database. Its users, settings, and share links are stored in/database/filebrowser.db. By default/databaseis a block PVC on a StatefulSet; only withstateful_pvc_enabled = falseis it a Cloud Storage bucket mounted via GCS FUSE (see below). - GCS FUSE vs. block PVC. With
stateful_pvc_enabled = true(the default) the workload becomes a StatefulSet with a persistent block PVC (default20Gi) mounted at/database, and the GCS FUSE volume is automatically disabled to avoid a double-mount at the same path. A block PVC gives SQLite proper POSIX file locking and is the more robust choice for a stateful file manager. - Single instance by design.
min_instance_count = max_instance_count = 1. SQLite does not tolerate concurrent writers — keep a single replica. - Default login is
admin/admin. Filebrowser seeds this on first boot; change it in the web UI immediately after deploy. - No Redis, no init job.
enable_redis = falseand nodb-initjob runs; the pod is ready as soon as the container starts. - Container port 80. Filebrowser serves plain HTTP/1.1 on port 80.
- Custom domain is on by default.
enable_custom_domain = trueandreserve_static_ip = true; supplyapplication_domainsto serve a hostname with a Google-managed certificate.
2. Google Cloud Services & How to Explore Them
All commands assume you have run
gcloud container clusters get-credentials <cluster> --region <region> --project <project>
and that PROJECT, REGION, and NAMESPACE are set. The namespace and other
identifiers are reported in the deployment Outputs.
A. GKE Autopilot — the Filebrowser workload
Filebrowser runs as a single-replica StatefulSet (stateful_pvc_enabled = true, the
default; a Deployment when it is false) scheduled on Autopilot, which bills for the CPU/memory
the pod actually requests.
- Console: Kubernetes Engine → Workloads → select the Filebrowser workload to see pods, revisions, and events. Kubernetes Engine → Services & Ingress shows the Service and any external IP.
- CLI:
kubectl get pods,svc -n "$NAMESPACE"
kubectl get statefulset,pvc -n "$NAMESPACE" # when stateful_pvc_enabled = true
kubectl logs -n "$NAMESPACE" deploy/<service-name> --tail=100
See App_GKE for how Autopilot, scaling, and the workload type (Deployment vs StatefulSet) are managed.
B. Cloud Storage / block PVC — persistent state
Filebrowser has no Cloud SQL database. Its embedded SQLite database
(/database/filebrowser.db) is stored on the /database mount:
-
Default — StatefulSet (
stateful_pvc_enabled = true): a block PersistentVolumeClaim (default20Gi, StorageClassstandard-rwo) mounted at/database; the GCS FUSE volume is disabled to avoid a double-mount. Keep it on: Filebrowser's database is bbolt, which writes at arbitrary offsets that GCS FUSE rejects. -
Deployment (
stateful_pvc_enabled = false): a dedicated Cloud Storage bucket mounted via GCS FUSE through the CSI driver. -
Console: Cloud Storage → Buckets; or Kubernetes Engine → Storage → PVCs.
-
CLI:
gcloud storage buckets list --project "$PROJECT" --filter="name~storage" # GCS FUSE mode
gcloud storage ls gs://<data-bucket>/filebrowser.db
kubectl get pvc -n "$NAMESPACE" # PVC mode
See App_GKE for CMEK options, GCS FUSE, and StatefulSet PVCs.
C. Secret Manager
Filebrowser generates no application secrets — there is no encryption key or JWT secret to manage, because all identity state lives in the SQLite database. Secret Manager is still used by the foundation for platform-managed secrets (e.g. CI/CD tokens if configured).
- Console: Security → Secret Manager.
- CLI:
gcloud secrets list --project "$PROJECT" --filter="name~filebrowser"
See App_GKE for the Secret Store CSI integration and rotation.
D. Networking & ingress
By default the Service is ClusterIP, with enable_custom_domain = true and
reserve_static_ip = true so an Ingress with a Google-managed certificate can serve
a supplied hostname on a stable IP. Without a custom domain the workload is reachable
in-cluster at http://<service>.<namespace>.svc.cluster.local.
- Console: Network services → Load balancing; VPC network → IP addresses.
- CLI:
kubectl get ingress,svc -n "$NAMESPACE"
gcloud compute addresses list --project "$PROJECT"
See App_GKE for custom domains, Cloud CDN, and static IP details.
E. Cloud Logging & Monitoring
Pod stdout/stderr flow to Cloud Logging; GKE metrics flow to Cloud Monitoring. Optional uptime checks and alert policies are available.
- Console: Logging → Logs Explorer; Monitoring → Dashboards / Alerting.
- CLI:
gcloud logging read 'resource.type="k8s_container" AND resource.labels.namespace_name="'"$NAMESPACE"'"' \
--project "$PROJECT" --limit 50
3. Filebrowser Application Behaviour
- No first-deploy database setup. There is no
db-initjob and no Cloud SQL instance. On first start the Filebrowser binary creates its SQLite database at/database/filebrowser.dbif it does not already exist and seeds the defaultadmin/adminuser. - State persistence. Users, settings, and share links live entirely in
/database/filebrowser.dbon the/databasemount (GCS FUSE bucket or block PVC), surviving restarts and redeploys.FB_ROOT = /srvis the file tree the app serves. - Default credentials must be changed. The seeded
admin/adminlogin is well-known. Log in and change the password (and ideally the username) in the web UI immediately after the first deploy. - Single-writer constraint. The embedded SQLite database does not support
concurrent writers. Keep
min_instance_count = max_instance_count = 1; a StatefulSet block PVC gives proper file locking but is still single-replica. - Health path. Startup and liveness probes target
/health— Filebrowser's unauthenticated health endpoint, which returns200as soon as the server is listening:kubectl exec -n "$NAMESPACE" deploy/<service-name> -- wget -qO- http://localhost:80/health - No Redis.
enable_redis = false; Filebrowser is a self-contained file manager with no queue or cache. The App_GKE default ofenable_redis = trueis explicitly overridden. - Custom-built image needs
imagePullPolicy = Always. The image is a thin wrapper built and mirrored into Artifact Registry; App_GKE setsimagePullPolicy = Alwaysfor custom/mirrored images so a rebuild-redeploy always pulls the fresh layer.
4. Configuration Variables
Variables are grouped exactly as they appear on the deployment platform. Only settings specific to or notable for Filebrowser are listed; every other input is inherited from App_GKE with its standard behaviour and defaults.
Group 1 — Project & Identity
| Variable | Default | Description |
|---|---|---|
project_id | (required) | Target Google Cloud project. |
region | us-central1 | Region for the workload and regional resources. |
Group 3 — Application Identity
| Variable | Default | Description |
|---|---|---|
application_name | filebrowser | Base name for resources. Do not change after first deploy. |
application_version | latest | Filebrowser image tag. latest resolves to the pinned v2.32.0 at build time; pin explicitly in production. |
Group 4 — Runtime & Scaling
| Variable | Default | Description |
|---|---|---|
deploy_application | true | Set false to provision infrastructure only. |
cpu_limit | 1000m | CPU per pod; Filebrowser is lightweight. |
memory_limit | 1Gi | Memory per pod; 256Mi is ample for the Go server. |
min_instance_count | 1 | Minimum replicas. Keep at 1 — SQLite is single-writer. |
max_instance_count | 1 | Keep at 1 to prevent concurrent SQLite writers. |
container_port | 80 | Filebrowser's HTTP/1.1 listener. |
enable_cloudsql_volume | false | Filebrowser has no Cloud SQL; leave false. |
enable_image_mirroring | true | Mirror the Filebrowser image into Artifact Registry. |
Group 6 — GKE Backend & Cluster
| Variable | Default | Description |
|---|---|---|
service_type | LoadBalancer | How the Kubernetes Service is exposed; front with an Ingress via enable_custom_domain. |
workload_type | null | Auto-resolves to StatefulSet when stateful_pvc_enabled = true, else Deployment. |
session_affinity | None | Single replica, so sticky routing is unnecessary. |
Group 7 — StatefulSet
| Variable | Default | Description |
|---|---|---|
stateful_pvc_enabled | true | Set true to store /database on a block PVC instead of GCS FUSE (recommended for SQLite file locking). |
stateful_pvc_size | 20Gi | Per-pod PVC storage size. |
stateful_pvc_mount_path | /database | Mount path — must match FB_DATABASE's directory. |
stateful_pvc_storage_class | standard-rwo | StorageClass (standard-rwo Balanced PD; premium-rwo for higher IOPS). |
Group 9 — Reliability Policies
| Variable | Default | Description |
|---|---|---|
enable_pod_disruption_budget | true | Protect availability during node upgrades. |
pdb_min_available | 1 | Minimum pods available during voluntary disruptions. |
Group 10 — Observability & Health
| Variable | Default | Description |
|---|---|---|
startup_probe | HTTP /health 15s delay | Startup probe; Filebrowser exposes /health once ready. |
liveness_probe | HTTP /health 30s delay | Liveness probe on the unauthenticated /health endpoint. |
uptime_check_config | {enabled=false, path="/health"} | Cloud Monitoring uptime check; disabled by default. |
alert_policies | [] | Optional metric alert policies. |
Group 13 — Filesystem (NFS)
| Variable | Default | Description |
|---|---|---|
enable_nfs | false | NFS is off by default; not needed for Filebrowser. |
nfs_mount_path | /mnt/nfs | Mount path inside the container. |
Group 14 — Cloud Storage & Artifact Registry
| Variable | Default | Description |
|---|---|---|
create_cloud_storage | true | Create the Filebrowser /database bucket (and any extra storage_buckets). |
storage_buckets | [] | Additional buckets to provision. |
gcs_volumes | [] | Extra GCS FUSE mounts. The /database bucket is added automatically (unless a PVC is used). |
manage_storage_kms_iam / enable_artifact_registry_cmek | false | CMEK options. |
Group 15 — Redis Cache & Queue
| Variable | Default | Description |
|---|---|---|
enable_redis | false | Inherited from App_GKE and not overridden by this module. Filebrowser uses no Redis, so set it to false when deploying — leaving it on wires an unused dependency. |
Group 16 — Database Backend
Not applicable — Filebrowser has no SQL database. database_password_length and
db_name / db_user are forwarded to the foundation only for compatibility;
database_type is fixed to NONE by Filebrowser_Common.
Group 19 — Custom Domain, Static IP & Networking
| Variable | Default | Description |
|---|---|---|
enable_custom_domain | true | Provision Ingress for custom hostnames + managed certificate. |
application_domains | [] | Hostnames to serve. |
reserve_static_ip | true | Stable external IP across redeploys. |
Group 20 — Identity-Aware Proxy (IAP)
| Variable | Default | Description |
|---|---|---|
enable_iap | false | Require Google sign-in in front of Filebrowser. |
iap_authorized_users / iap_authorized_groups | [] | Who may access. |
iap_oauth_client_id / iap_oauth_client_secret | "" | Required when IAP is enabled (sensitive). |
Group 22 — VPC Service Controls & Audit Logging
| Variable | Default | Description |
|---|---|---|
enable_vpc_sc | false | Enforce a VPC-SC perimeter (requires organization_id). |
vpc_cidr_ranges / vpc_sc_dry_run | (set) | Access level CIDRs / dry-run mode. |
enable_audit_logging | false | Detailed Cloud Audit Logs. |
All other inputs follow standard App_GKE behaviour.
5. Outputs
These values are returned on a successful deployment and are the quickest way to locate and explore the running resources.
| Output | Description |
|---|---|
service_name | Kubernetes Service name. |
namespace | Namespace the workload runs in. |
service_cluster_ip | In-cluster ClusterIP. |
stage_service_cluster_ips | Map of ClusterIPs for stage-specific services. |
service_external_ip | External LoadBalancer IP (when a static IP is reserved). |
service_url | URL to reach Filebrowser. |
storage_buckets | Created Cloud Storage buckets (includes the /database bucket in GCS FUSE mode). |
network_name / network_exists / regions | VPC network, presence, available regions. |
container_image / container_registry | Deployed image and Artifact Registry repo. |
monitoring_enabled / monitoring_notification_channels | Monitoring status and channels. |
initialization_jobs | Names of any init jobs (empty by default). |
statefulset_name | Name of the StatefulSet (when stateful_pvc_enabled = true). |
deployment_id / tenant_id / resource_prefix | Naming identifiers. |
project_id / project_number | Project identifiers. |
cicd_enabled / cicd_configuration | CI/CD status and details (repo, trigger, registry). |
github_repository_url / github_repository_owner / github_repository_name | CI/CD GitHub details. |
artifact_registry_repository / cloudbuild_trigger_name / cloudbuild_trigger_id | Registry and build trigger. |
kubernetes_ready | Whether the cluster/workload is ready. |
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_GKE foundation engine, which validates values and combinations at plan time — IAP with no OAuth credentials,
min_instance_count > max_instance_count,workload_type = Deploymentalongsidestateful_pvc_enabled = true, ResourceQuota memory values without binary unit suffixes. 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 |
|---|---|---|---|
/database volume (bucket or PVC) | Never delete | Critical | The embedded SQLite DB lives here; deleting it destroys all users, settings, and share links. |
admin / admin (seeded login) | Change on first login | Critical | Leaving the default credential lets anyone who can reach the service take full control. |
max_instance_count | 1 | High | >1 puts concurrent writers on the single SQLite database, corrupting it. |
stateful_pvc_mount_path | /database | High | Must match FB_DATABASE's directory; a mismatch stores the DB on ephemeral disk and loses state on restart. |
stateful_pvc_enabled + enable_gcs_storage_volume | Let Common disable GCS FUSE | High | Both at /database double-mount; Common auto-sets enable_gcs_storage_volume = false when the PVC is on — do not force both. |
container_port | 80 | High | Filebrowser listens on 80; a different port makes the startup probe fail and the pod never becomes Ready. |
startup_probe / liveness_probe path | /health | High | Pointing probes at an authenticated path returns 401/403 and the pod never goes Ready. |
enable_cloudsql_volume | false | Medium | Filebrowser has no Cloud SQL; enabling adds a useless Auth Proxy sidecar. |
enable_redis | true | Medium | Filebrowser has no Redis, but the App_GKE default of true is inherited unchanged — set it to false explicitly, or an unused dependency is wired in. |
enable_iap | credentials required | High | Enabling IAP without iap_oauth_client_id/secret silently exposes the service unauthenticated (blocked by a plan-time guard). |
application_version | pin in production | Medium | latest resolves to a pinned v2.32.0 at build time; pin explicitly to control upgrades. |
For the foundation behaviour referenced throughout — IAM and Workload Identity, autoscaling, ingress and certificates, CI/CD, Cloud Armor, IAP, Binary Authorization, VPC-SC, backups, and image mirroring — see App_GKE. Filebrowser-specific application configuration shared with the Cloud Run variant is described in Filebrowser_Common.
Related guides
- Hands-on lab: Filebrowser on GKE Autopilot — deploy it step by step, with the console screens and commands at each stage.
- Filebrowser Common — Shared Application Configuration — the configuration shared by both deployment targets.
- Deployed alongside Kopia on GKE Autopilot, UrBackup on GKE Autopilot in the Backup & Disaster Recovery solution.
Need RAD to do something it does not do yet? Request it on the roadmap, or vote on what is already there.