Superset GKE Module — Configuration Guide

This guide describes every configuration variable available in the Superset_GKE module. Superset_GKE is a wrapper module that combines the generic App_GKE infrastructure module with the Superset_Common shared application configuration to deploy Apache Superset on Google Kubernetes Engine (GKE) Autopilot.

Most configuration options in Superset GKE map directly to the same options in App GKE. Where a variable is identical in behaviour, this guide references the App GKE guide rather than repeating the documentation.

Note: Variables marked as platform-managed are set and maintained by the platform. You do not normally need to change them.

---

Standard Configuration Reference

Configuration Area	App GKE.md Section	Superset-Specific Notes
Module Metadata & Configuration	§1 Module Overview	Superset-specific module_description and module_services defaults are pre-set.
Project & Identity	§2 IAM & Access Control	Identical.
Application Identity	§3.A Compute (GKE Autopilot)	Superset-specific defaults; see Group 2.
Runtime & Scaling	§3.A Compute (GKE Autopilot)	Superset-specific defaults; see Group 3.
Environment Variables & Secrets	§3 Core Service Configuration	SUPERSET_SECRET_KEY auto-injected; see Group 5.
Initialization Jobs & CronJobs	§3.E Initialization Jobs & CronJobs	Two-phase db-init + app-init; see Group 8.
Storage — GCS	§3.C Storage	superset-data GCS bucket provisioned automatically.
Database Configuration	§3.B Database (Cloud SQL)	PostgreSQL 15; see Group 11.
Observability & Health Checks	§3.A Compute (GKE Autopilot)	Probes target /health; see Group 13.
Cloud Armor WAF	§4.A Cloud Armor WAF	Identical.
Identity-Aware Proxy	§4.B IAP	Useful for restricting Superset to internal users.
Traffic & Ingress	§5 Traffic & Ingress	session_affinity = "ClientIP" recommended; see Group 17.
Redis Cache	§8.A Redis / Memorystore	enable_redis = false by default; see Group 14.

---

How Superset GKE Relates to App GKE

1. SUPERSET_SECRET_KEY is auto-generated. Superset Common creates a 50-character random key in Secret Manager. This key signs Flask sessions — changing it invalidates all active sessions.
2. Two-phase initialisation. Superset Common provides db-init (database creation) and app-init (Superset schema migration and admin creation) jobs that run automatically on first deploy.
3. superset-data GCS bucket is provisioned automatically.
4. PostgreSQL 15 is the supported database. database_type defaults to "POSTGRES_15".
5. Session affinity is recommended. session_affinity = "ClientIP" is the default to ensure Superset's stateful session handling works correctly across multiple pods.
6. Health probes target /health. Superset exposes a /health endpoint that returns HTTP 200 when the Gunicorn worker pool is ready.

---

Group 0: Module Metadata & Configuration

Identical to App_GKE. See App_GKE §1.

Superset-specific defaults:

Variable	Superset GKE Default	Notes
credit_cost	150	GKE deployments cost more credits than Cloud Run.

---

Group 1: Project & Identity

Identical to App_GKE. See App_GKE §2.

---

Group 2: Application Identity

Superset-specific defaults:

Variable	Superset GKE Default	App GKE Default	Notes
application_name	"superset"	"gkeapp"	Base name for all resources. Do not change after deployment.
display_name	"Superset"	(not in App GKE)	Human-readable name.
description	"Apache Superset data visualisation platform"	(not in App GKE)	Deployment description.
application_version	"latest"	"1.0.0"	Superset release version.

---

Group 3: Runtime & Scaling

Superset-specific defaults:

Variable	Superset GKE Default	App GKE Default	Notes
container_port	8088	8080	Superset's Gunicorn port.
cpu_limit	"2000m"	"1000m"	Python query execution requires 2 vCPU.
memory_limit	"2Gi"	"512Mi"	Query result caching and Pandas DataFrames require 2 Gi.
min_instance_count	1	1	Always one warm pod.
max_instance_count	3	3	Multiple concurrent users.
enable_cloudsql_volume	true	true	Cloud SQL Auth Proxy sidecar.
timeout_seconds	600	300	Extended for long-running queries.

---

Group 4: Access & Networking

Identical to App_GKE. See App_GKE §4.

Variable	Default	Description
enable_iap	false	Identity-Aware Proxy. Restricts Superset to authenticated Google users.
iap_authorized_users	[]	Users/service accounts granted IAP access.
iap_authorized_groups	[]	Google Groups granted IAP access.
enable_custom_domain	false	Custom domain with SSL.
application_domains	[]	Custom domain names.
reserve_static_ip	true	Reserves a Global Static IP.
enable_cloud_armor	false	Cloud Armor WAF.
admin_ip_ranges	[]	WAF-exempt CIDR ranges.
enable_vpc_sc	false	VPC Service Controls.

---

Group 5: Environment Variables & Secrets

SUPERSET_SECRET_KEY is injected automatically from Superset Common. The value is a 50-character random string stored in Secret Manager — do not rotate it without coordinating session invalidation.

Variable	Default	Description
environment_variables	{}	Additional env vars.
secret_environment_variables	{}	Additional Secret Manager references. SUPERSET_SECRET_KEY is auto-injected.

---

Group 6: Backup & Maintenance

Identical to App_GKE. See App_GKE §3.B.

Variable	Default	Notes
backup_schedule	"0 2 * * *"	Daily at 02:00 UTC.
backup_retention_days	7	7-day retention.
enable_backup_import	false	One-time restore on deploy.

---

Group 7: CI/CD & GitHub Integration

Identical to App_GKE. See App_GKE §6.

---

Group 8: Jobs & Scheduled Tasks

Superset default two-phase init pipeline:

Job	Image	Purpose	Depends On	Timeout	execute_on_apply
db-init	postgres:15-alpine	Create Superset PostgreSQL database and user	—	600s	true
app-init	(Superset app image)	Run superset db upgrade + superset fab create-admin	db-init	1800s	true

The app-init job runs the Superset container image itself and executes the combined database upgrade and initial admin user creation. The 30-minute timeout accommodates schema migrations on large or complex database setups.

Override initialization_jobs with a non-empty list to replace this default pipeline.

---

Group 11: Database Configuration

Superset-specific defaults:

Variable	Superset GKE Default	App GKE Default	Notes
database_type	"POSTGRES_15"	"POSTGRES"	Superset requires PostgreSQL.
db_name	"superset_db"	(not in App GKE)	Database name passed to Superset Common.
db_user	"superset_user"	(not in App GKE)	Database user passed to Superset Common.

Automatic password rotation:

Variable	Default	Description
enable_auto_password_rotation	false	Automated rotation.
rotation_propagation_delay_sec	90	Restart delay after rotation.

---

Group 13: Observability & Health

Superset exposes /health as its dedicated health endpoint.

Startup probe:

Field	Superset Default	Notes
path	"/health"	Superset's health endpoint.
initial_delay_seconds	60	Gunicorn worker pool initialisation takes time.
failure_threshold	12	Allows up to 180s total startup tolerance.

Liveness probe:

Field	Superset Default
path	"/health"
initial_delay_seconds	30

Override recommended: startup_probe_config and health_check_config default to path = "/healthz". Override both to path = "/health" for Superset.

---

Group 14: Reliability Policies

Identical to App_GKE. See App_GKE §7.

---

Group 14: Redis Cache

Redis is disabled by default but recommended for production multi-user deployments. Redis serves as the Celery broker for async queries and the result backend for caching.

Variable	Default	Description
enable_redis	false	Enables Redis. Recommended for production.
redis_host	""	Redis hostname or IP.
redis_port	"6379"	Redis port (string in Superset GKE, unlike number in Superset CloudRun).
redis_auth	""	Redis AUTH password. Sensitive.

---

Group 17: GKE Backend Configuration

Superset-specific defaults:

Variable	Superset GKE Default	App GKE Default	Notes
session_affinity	"ClientIP"	"None"	Ensures Superset sessions are consistently routed to the same pod. Required for reliable login behaviour.
workload_type	null	null	Defaults to Deployment.
service_type	"LoadBalancer"	"LoadBalancer"	External load balancer.

---

Module Outputs

Output	Description
service_name	Kubernetes service name.
service_url	Service URL.
service_external_ip	External load balancer IP.
project_id	GCP project ID.
deployment_id	Deployment ID suffix.
namespace	Kubernetes namespace.
database_instance_name	Cloud SQL PostgreSQL 15 instance name.
database_name	Application database name.
database_user	Application database user.
database_password_secret	Secret Manager secret for the database password.
storage_buckets	Created GCS storage buckets.
container_image	Container image used.
kubernetes_ready	true when Kubernetes resources are deployed.

Configuration Pitfalls & Sensible Defaults

Risk levels: Critical (data loss, full outage, security breach) — High (service unavailable or significant degradation) — Medium (degraded function or increased cost) — Low (minor impact).

Variable	Sensible Default	Risk	Consequence of Incorrect Value
SUPERSET_SECRET_KEY (via Secret Manager)	Auto-generated 50-char random string	Critical	If this value is changed after the first deployment, all existing user sessions are immediately invalidated and all encrypted data source credentials stored in Superset's PostgreSQL schema become permanently unreadable. Treat as immutable after first deploy.
container_resources.memory_limit	"2Gi"	High	Under 1 Gi gunicorn workers are OOM-killed during query execution. "2Gi" is the minimum; "4Gi" is recommended for production. On GKE Autopilot, mem_request also drives node provisioning — set close to memory_limit.
container_resources.cpu_limit	"2000m"	High	Superset migrations (run by the app-init job) and gunicorn startup require significant CPU. Under 1000m the init job may time out in its 1800 s window.
container_resources.mem_request	null (defaults to limit)	Medium	On GKE Autopilot, setting mem_request far below memory_limit leads to burstable scheduling and possible OOM eviction under memory pressure.
enable_redis	false	High	Without Redis, Celery workers have no broker or result backend. Async query execution, cache warming, and scheduled reports are all non-functional. For GKE production deployments, always set enable_redis = true.
redis_host	null	High	Required when enable_redis = true. An empty value causes all Celery workers to fail to connect on pod startup, making async queries permanently unavailable.
SUPERSET_PORT	"8088" (injected)	High	Must match container_port. Changing one without the other breaks all routing and health checks.
application_database_name	"superset"	High	Immutable after the db-init job has run. Changing orphans the entire Superset schema.
application_database_user	"superset"	High	Immutable after the db-init job has run. Renaming requires manual Cloud SQL intervention.
application_version	"latest"	Medium	Pinning to a specific version prevents uncontrolled upgrades that may introduce breaking API changes. Always test upgrades in staging.
min_instance_count	1	High	Scale-to-zero terminates Celery workers; async queries submitted while the pod is cold are lost. Superset has a 30–60 s startup time.
max_instance_count	(check your setting)	Medium	Multiple replicas share PostgreSQL but require Redis as a shared Celery result backend. Without Redis, async results are only accessible to the instance that executed the query.
quota_memory_requests / quota_memory_limits	"4Gi" / "8Gi"	High	GKE-specific: must use binary suffixes (Gi, Mi). Bare integers (e.g., "4") are treated as bytes by Kubernetes and block all pod scheduling.
stateful_pvc_enabled	false	Medium	Superset does not need persistent volumes — state is in PostgreSQL and Redis. Enabling adds unnecessary StatefulSet complexity.
pdb_min_available	"1"	Medium	Setting to "0" allows all pods to be evicted simultaneously during node upgrades, causing a full Superset outage.
enable_iap	false	High	Without IAP the Superset login form is reachable from the load-balancer IP. Always enable IAP or configure Kubernetes network policies for production.
startup_probe_config.failure_threshold	30	High	Reducing below 15 causes GKE to kill pods before Superset completes db migrations and starts gunicorn.
backup_schedule	"0 2 * * *"	Medium	Disabling automated backups leaves all dashboards, charts, and RLS rules unprotected.
db_tier	"db-f1-micro" (Common default)	Medium	Insufficient for production Superset workloads. Override to at least "db-custom-2-7680" in production environments.
SUPERSET_LOAD_EXAMPLES	"no" (injected)	Medium	Overriding to "yes" populates the workspace with demo data on every startup and significantly increases init time.