Jellyfin on GKE Autopilot

Jellyfin is a free, open-source (GPLv2) self-hosted media server for streaming your own movies, TV shows, music, photos, and live TV. Written in .NET/C# and maintained as a community fork of Emby, it has no tracking, no ads, and no premium tier. This module deploys Jellyfin 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 Jellyfin 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.

---

1. Overview

Jellyfin runs as a stateful .NET workload. On GKE this is the recommended home for a real media library: a StatefulSet backed by a real block PVC at /config gives correct POSIX filesystem semantics for SQLite and the transcode cache. The deployment wires together a focused set of Google Cloud services:

Capability	Google Cloud service	Notes
Compute	GKE Autopilot	.NET StatefulSet pod, 1 vCPU / 1 GiB by default
Persistence	Persistent Disk (block PVC)	/config on a per-pod PVC — the recommended store for SQLite + transcode cache
Database	Internal SQLite (embedded)	No Cloud SQL — Jellyfin keeps all state in SQLite files under /config
Object / file storage	Cloud Storage (GCS FUSE) / Filestore (NFS)	Optional, for large media libraries
Secrets	Secret Manager	Optional auto-generated API key; no mandatory cryptographic secrets
Ingress	Cloud Load Balancing	ClusterIP by default; optional custom domain + managed certificate

Sensible defaults worth knowing up front:

• There is no external database. Jellyfin stores its entire state — the SQLite library and playback databases, configuration XML, cached metadata and artwork, plugins, transcode cache, and logs — under /config. No Cloud SQL instance, no db-init job, and no Redis is used (database_type = NONE; the foundation Redis variables are inert for Jellyfin).
• A block PVC at /config is the best fit. stateful_pvc_enabled = true resolves the workload to a StatefulSet with a per-pod PVC mounted at /config, and the GCS storage volume auto-disables to avoid a double mount. Real block storage gives the correct filesystem semantics SQLite and the transcode cache need — the recommended configuration for a media server.
• The container listens on port 8096. Jellyfin's web/API port is set by Jellyfin_Common. The web UI and first-run setup wizard are served at /web (and /); GET /health returns Healthy (200, unauthenticated).
• There are no default credentials. On first access the setup wizard creates the administrator account and adds media libraries. Nothing is usable until then.
• Single replica. min_instance_count = 1 / max_instance_count = 1 — one shared SQLite library on one volume. Do not run multiple replicas; concurrent writers against one SQLite file corrupt the library.
• NFS is optional, for large libraries. enable_nfs = false by default. Enable it to mount a shared Filestore volume for a large media collection that outgrows a single PVC.
• API-key auth is optional and off by default. enable_api_key = false. Primary authentication is the wizard-created admin account; per-application API keys are created in-app under Dashboard → API Keys.

GKE vs Cloud Run — GKE is the production media server. GKE (this module) runs Jellyfin as a StatefulSet with a real block PVC at /config, giving true POSIX semantics for SQLite and the transcode cache, plus optional NFS for large media libraries — the recommended choice for a real, multi-user, transcoding media server. Jellyfin_CloudRun mounts /config from a GCS bucket over FUSE; it is simpler and cheaper for a demo or a small personal library, but FUSE latency and the per-request timeout model make it a poor fit for live transcoding or busy streaming.

---

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 Jellyfin workload

Jellyfin pods are scheduled on Autopilot, which bills for the CPU/memory the pods actually request. With a block PVC the workload is a StatefulSet with a stable pod identity and orderly restarts.

• Console: Kubernetes Engine → Workloads → select the Jellyfin workload to see pods, events, and the StatefulSet. Kubernetes Engine → Services & Ingress shows the external IP (when exposed).
• CLI:

  kubectl get pods,svc,statefulset,pvc -n "$NAMESPACE"
  kubectl logs -n "$NAMESPACE" statefulset/<service-name> --tail=100
  kubectl describe pod -n "$NAMESPACE" <pod-name>

See App_GKE for how Autopilot, scaling, and the workload type (Deployment vs StatefulSet) are managed.

B. Persistent configuration store (SQLite on the PVC)

Jellyfin has no external database. Its entire state — the SQLite library and playback databases, configuration XML, cached metadata and artwork, installed plugins, the transcode cache, and logs — lives under /config (JELLYFIN_CONFIG_DIR = /config). There is no Cloud SQL instance, no Auth Proxy, and no initialization Job to create a schema; Jellyfin creates and migrates its own SQLite databases on first start.

On GKE, /config is backed by a per-pod block PVC (stateful_pvc_enabled = true, mounted at /config), which is the recommended store because SQLite and the transcode cache need true POSIX filesystem semantics that object storage cannot provide.

• Inspect the PVC and its bound disk:

  kubectl get pvc -n "$NAMESPACE"
  kubectl describe pvc -n "$NAMESPACE" <pvc-name>

C. Cloud Storage & NFS (optional, for large libraries)

The auto-provisioned Cloud Storage bucket (name suffix storage, STANDARD, force_destroy = true, versioning off, public_access_prevention = enforced) is available for additional GCS FUSE mounts. When a block PVC is enabled the GCS storage volume is auto-disabled for /config to avoid a double mount. For large media collections, enable NFS (enable_nfs = true) to mount a shared Filestore volume.

• Console: Cloud Storage → Buckets; Filestore → Instances.
• CLI:

  gcloud storage buckets list --project "$PROJECT"
  gcloud filestore instances list --project "$PROJECT"

See App_GKE for CMEK options and the GCS FUSE CSI driver.

D. First-run setup & the media library

On first access Jellyfin serves an interactive setup wizard at /web (and /) that creates the administrator account, sets the preferred language, and lets you add media libraries (Movies, TV, Music, Photos). Nothing is authenticated or usable until you complete the wizard — there are no default credentials.

Media libraries point at paths inside the container: the block PVC at /config, an optional NFS mount for large collections, or additional GCS FUSE volumes.

• Reach the wizard / web UI:

  kubectl get svc,ingress -n "$NAMESPACE"      # external IP / hostname
  kubectl port-forward -n "$NAMESPACE" statefulset/<service-name> 8096:8096
  # then open http://localhost:8096/web

E. Secret Manager & the optional API key

Jellyfin requires no mandatory cryptographic secrets — there is no encryption key, JWT, or master password to manage. When enable_api_key = true, the module generates a 32-character random value, stores it in Secret Manager as secret-<prefix>-<app>-api-key (surfaced as the jellyfin_api_key_secret_id output), and injects it so external callers can authenticate programmatically. In day-to-day use, API keys are created and revoked in-app under Dashboard → API Keys; primary auth remains the wizard admin account.

• Console: Security → Secret Manager.
• CLI:

  gcloud secrets list --project "$PROJECT"
  gcloud secrets versions access latest --secret=<secret-name> --project "$PROJECT"

See App_GKE for the Secret Store CSI integration and rotation.

F. Networking & ingress

By default service_type = ClusterIP, keeping the media server in-cluster. Set service_type = LoadBalancer for an external IP, or enable a custom domain with a Google-managed certificate via the Kubernetes Gateway API. A static IP can be reserved so the address survives redeploys.

• 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.

G. 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. Jellyfin Application Behaviour

• No initialization Job. Jellyfin needs no db-init step — it creates and migrates its own SQLite databases under /config the first time it starts. Leave initialization_jobs empty unless you have custom data-loading tasks.
• First-run wizard creates the admin. The /web setup wizard walks you through creating the administrator account and adding libraries. Until it is completed the server has no users and no content.
• /config is the single source of truth — persist it. All library state is on the block PVC. Deleting the PVC wipes the library, plugins, and users. The StatefulSet keeps the PVC bound to the pod identity across restarts.
• Custom image is a thin wrapper. The Dockerfile is ARG JELLYFIN_VERSION=10.10.3 / FROM jellyfin/jellyfin:${JELLYFIN_VERSION}, so the Foundation mirrors it into Artifact Registry (enable_image_mirroring = true) and sets imagePullPolicy = Always for the mirrored image. application_version = "latest" resolves to the pinned 10.10.3 via the app-specific JELLYFIN_VERSION build arg — it is not overwritten by the Foundation's generic APP_VERSION injection.
• fsGroup for a group-writable PVC. Jellyfin runs as UID 1000 / GID 2000; stateful_fs_group = 3000 (the Jellyfin Helm chart default) ensures the PVC is group-writable.
• Health path. Startup and liveness probes target GET /health, which returns Healthy (200) without authentication once the server is ready. The startup probe allows a 15-second initial delay with a generous retry window; the liveness probe polls every 30 seconds.
• Transcoding is CPU-heavy and GPU-less. Autopilot pods have no GPU, so prefer direct-play clients. Size cpu_limit up for live transcoding and memory_limit up for large libraries.

---

4. Configuration Variables

Variables are grouped exactly as they appear on the deployment platform. Only settings specific to or notable for Jellyfin 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 2 — Deployment Environment

Variable	Default	Description
tenant_deployment_id	demo	Short suffix that makes resource names unique per environment.
support_users	[]	Emails granted project access and monitoring alerts.
resource_labels	{}	Labels applied to all resources for cost/ownership tracking.

Group 3 — Application Identity

Variable	Default	Description
application_name	jellyfin	Base name for resources. Do not change after first deploy.
application_display_name	Jellyfin Media Server	Human-readable name shown in the platform UI.
application_version	latest	Jellyfin image tag; latest pins to 10.10.3 via the JELLYFIN_VERSION build arg.
enable_api_key	false	Generate a random API key in Secret Manager. Recommended when reachable outside the namespace.

Group 4 — Runtime & Scaling

Variable	Default	Description
deploy_application	true	Set false to provision infrastructure only.
cpu_limit	1000m	CPU per pod; raise for live transcoding.
memory_limit	1Gi	Memory per pod; raise for large libraries.
min_instance_count	1	Minimum replicas; keep at 1 (single shared library).
max_instance_count	1	Keep at 1. One shared SQLite library on one volume — never run multiple replicas.
container_port	8096	Jellyfin's web/API port (set by Jellyfin_Common; not forwarded to App_GKE).
timeout_seconds	300	Maximum request duration (0–3600 seconds).
enable_cloudsql_volume	false	Jellyfin has no Cloud SQL — leave false.
enable_image_mirroring	true	Mirror jellyfin/jellyfin into Artifact Registry.
enable_vertical_pod_autoscaling	false	VPA optimises requests; disables HPA when on.

Group 5 — Environment Variables & Secrets

Variable	Default	Description
environment_variables	{}	Extra non-secret settings for the Jellyfin container.
secret_environment_variables	{}	Map of env var → Secret Manager secret name.
secret_propagation_delay	30	Seconds to wait after secret creation before proceeding.
secret_rotation_period	2592000s	Secret Manager rotation notification frequency.

Group 6 — GKE Backend & Cluster

Variable	Default	Description
service_type	ClusterIP	How the Kubernetes Service is exposed; LoadBalancer for external access.
workload_type	null → StatefulSet	Resolves to StatefulSet when stateful_pvc_enabled = true.
session_affinity	None	Session affinity mode for the Service.
namespace_name	""	Auto-generated from application_name + tenant_deployment_id when empty.
network_tags	["nfsserver"]	nfsserver is required when enable_nfs = true.
termination_grace_period_seconds	60	Seconds after SIGTERM before SIGKILL — allows in-flight writes to flush.
enable_network_segmentation	false	Create Kubernetes NetworkPolicy resources.

Group 7 — StatefulSet

Variable	Default	Description
stateful_pvc_enabled	null	Enable the PVC template. Recommended true for Jellyfin — auto-resolves to StatefulSet.
stateful_pvc_size	20Gi	Per-pod PVC size; size to hold /config (SQLite, metadata, transcode cache).
stateful_pvc_mount_path	/config	Container mount path for the PVC (Jellyfin's config/persistence dir).
stateful_pvc_storage_class	standard-rwo	Balanced PD; use premium-rwo for higher IOPS.
stateful_headless_service	null	Headless Service for stable pod DNS names.
stateful_pod_management_policy	null → OrderedReady	Safe ordered restarts for Jellyfin.
stateful_update_strategy	null → RollingUpdate	Update strategy.
stateful_fs_group	3000	Pod fsGroup so the PVC is group-writable (Jellyfin UID 1000 / GID 2000).

Group 8 — Resource Quota

enable_resource_quota (false) plus quota_cpu_requests / quota_cpu_limits / quota_memory_requests / quota_memory_limits / quota_max_pods / quota_max_services / quota_max_pvcs — namespace ResourceQuota. The quota *_requests / *_limits values are not forwarded in this module and are inert; memory values, if used elsewhere, must carry binary unit suffixes (4Gi, 8192Mi).

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; /health returns 200 once ready.
liveness_probe	HTTP /health 30s delay	Liveness probe.
startup_probe_config	HTTP /health	App_GKE-level infrastructure startup probe.
health_check_config	HTTP /health	App_GKE-level liveness probe.
uptime_check_config	{ enabled=false, path="/health" }	Optional Cloud Monitoring uptime check.
alert_policies	[]	Optional metric alert policies.

Group 11 — Jobs & Scheduled Tasks

Variable	Default	Description
initialization_jobs	[]	Jellyfin needs no init job; provide only for custom data-loading tasks.
cron_jobs	[]	Scheduled Kubernetes CronJobs (e.g. maintenance tasks).
additional_services	[]	Sidecar or helper services deployed alongside Jellyfin.

Group 12 — CI/CD & GitHub Integration

Standard App_GKE Cloud Build / Cloud Deploy integration — see App_GKE. Key inputs: enable_cicd_trigger, github_repository_url, github_token, enable_cloud_deploy, enable_binary_authorization.

Group 13 — Filesystem (NFS)

Variable	Default	Description
enable_nfs	false	Provision Cloud Filestore (NFS); enable for large shared media libraries.
nfs_mount_path	/mnt/nfs	Mount path inside the container.
nfs_volume_name	nfs-data-volume	Volume name for the NFS mount.

Group 14 — Cloud Storage & Artifact Registry

Variable	Default	Description
create_cloud_storage	true	Provision the Jellyfin storage bucket and any extras.
storage_buckets	[]	Additional buckets to provision.
gcs_volumes	[]	Additional GCS FUSE volume mounts via the CSI driver.
manage_storage_kms_iam / enable_artifact_registry_cmek	false	CMEK options.
max_images_to_retain / delete_untagged_images / image_retention_days	7 / true / 30	Artifact Registry cleanup policy.

Group 16 — Database Configuration

database_type (NONE), database_password_length, application_database_name (jellyfindb), application_database_user (jellyfinuser), enable_mysql_plugins, enable_postgres_extensions, db_* / db_*_env_var_name — all inert for Jellyfin (no SQL database); retained and forwarded for foundation compatibility.

Group 15 — Redis (forwarded for foundation compatibility)

enable_redis, redis_host, redis_port, redis_auth — not applicable to Jellyfin, which uses no cache or queue. Forwarded to the foundation only for compatibility; leave at defaults.

Group 17 — Backup & Maintenance

Variable	Default	Description
backup_schedule	0 2 * * *	Automated backup cron (UTC) of the /config volume.
backup_retention_days	7	Retention; raise to 30–90 for production.
enable_backup_import / backup_source / backup_uri / backup_format	restore options	Restore a /config snapshot on deploy (tar default).

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.
static_ip_name	""	Auto-generated when empty.

Group 20 — Identity-Aware Proxy (IAP)

Warning: Enabling IAP requires Google identity authentication for all inbound requests. Requires enable_custom_domain or enable_cdn to be true.

Variable	Default	Description
enable_iap	false	Require Google sign-in in front of Jellyfin.
iap_authorized_users / iap_authorized_groups	[]	Who may access.
iap_oauth_client_id / iap_oauth_client_secret	""	Required when IAP is enabled (sensitive).

Group 21 — Cloud Armor & CDN

Variable	Default	Description
enable_cloud_armor	false	Attach a Cloud Armor (WAF) policy to the Ingress backend.
admin_ip_ranges	[]	CIDRs allowed privileged access.
cloud_armor_policy_name	default-waf-policy	Policy name.
enable_cdn	false	Enable Cloud CDN on the GKE Ingress backend.

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	[] / true	Access level CIDRs / dry-run mode.
enable_audit_logging	false	Detailed Cloud Audit Logs.

---

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 Jellyfin.
jellyfin_api_key_secret_id	Secret Manager secret ID for the API key (empty when enable_api_key = false).
storage_buckets	Created Cloud Storage buckets.
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 setup jobs (empty for a default Jellyfin deploy).
statefulset_name	Name of the StatefulSet.
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 — a Deployment workload alongside stateful_pvc_enabled = true, IAP with no authorized identities, quota_memory_* without binary unit suffixes, an out-of-range timeout_seconds/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
/config PVC	Never delete	Critical	The PVC holds the SQLite library, users, and metadata; deleting it wipes the entire server.
stateful_pvc_enabled	true	Critical	Without a persistent PVC, /config is ephemeral and the library is lost on every pod restart.
max_instance_count	1	Critical	Multiple replicas write to one SQLite library and corrupt it.
workload_type vs stateful_pvc_enabled	Leave workload_type unset	Critical	Deployment + stateful_pvc_enabled = true fails at plan time; leave unset to auto-resolve to StatefulSet.
enable_backup_import	false unless restoring	Critical	Enabling without a valid backup_uri fails the import job.
stateful_pvc_size	Size to library	High	An undersized PVC fills up during metadata/transcode caching and stalls the server.
stateful_fs_group	3000	High	A wrong fsGroup leaves the PVC non-writable by Jellyfin (UID 1000 / GID 2000) — startup fails.
memory_limit	1Gi (raise for large libraries)	High	Too little memory OOM-kills the pod while scanning or transcoding a large library.
cpu_limit	1000m (raise for transcoding)	High	Live transcoding (no GPU) saturates CPU; prefer direct-play clients.
min_instance_count	1	High	GKE requires min ≥ 1; the validation guard rejects invalid values.
quota_memory_requests / _limits	binary units (4Gi, 8192Mi)	Critical	Bare integers are bytes and block all pod scheduling in the namespace.
enable_api_key	true when externally reachable	Medium	Without it, the API surface relies solely on session auth once exposed.
enable_pod_disruption_budget	true	Medium	Disabling allows GKE to evict the single pod during maintenance, interrupting streams.
backup_retention_days	7 (raise for prod)	Medium	Too short to recover an older library snapshot.

---

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. Jellyfin-specific application configuration shared with the Cloud Run variant is described in Jellyfin_Common. For a guided walkthrough, see the Jellyfin_GKE lab.