Which template should I pick?

The two-minute answer

Most tenants pick these first:

PostgreSQL — the default relational database. Install this early, point your apps at it, don't overthink it.
Redis — for caches, queues, rate limits, session stores. Install when your first app that needs it goes in.
MinIO — S3-compatible object storage. Install when you need to store files (uploads, generated PDFs, backups) without paying S3.
Grafana — dashboards for the platform's built-in Prometheus. Install once, then every future app gets observability for free.

Everything else is need-driven.

By what you're trying to do

I need to store structured data

→ PostgreSQL. It's boring, correct, and battle-tested. Skip "NoSQL first" instincts unless you already know why you need one.

I need a cache or a queue

→ Redis. Sub-millisecond gets, pub/sub, streams, and it's the queue backend for almost every job runner. If you outgrow it, you'll know.

I need to store files (uploads, backups, generated blobs)

→ MinIO. S3-compatible so your app code stays portable; when you move to real S3 later, it's a URL change.

I need to see graphs of what's happening

→ Grafana. Talks to the Prometheus instance the platform ships with. Two clicks to a dashboard for CPU, memory, and per-pod metrics.

I need to watch for downtime and get pinged

→ Uptime Kuma. HTTP/TCP/DNS/ping checks, status pages, notifications to Slack/Discord/email.

I need to automate a workflow (webhooks, integrations, scheduled jobs)

→ n8n. Visual editor, hundreds of nodes, self-hosted. Think Zapier but you own the box.

→ Ghost. Editorial-first, has memberships and Stripe integration built in. If you need a general-purpose CMS with plugins, look elsewhere; Ghost is opinionated on purpose.

→ Plausible. Cookieless, GDPR-friendly, tiny script. Doesn't do per-user tracking (that's the point).

I need full-text search inside my app

→ Meilisearch. Fast, typo-tolerant, easy to run. Elasticsearch is overkill until it isn't.

I need to test emails locally

→ MailHog. Catches SMTP, shows the emails in a web UI. Only for dev namespaces — do not run in prod.

Things to know before installing

Placeholder passwords. Every template ships with change-me-in-secret as the default password. Replace it in the Values YAML before submitting, or wire it to an ExternalSecret (see the custom chart guide for the pattern).

Storage sizes. Templates default to modest sizes (2–20 GiB). If you know you'll need more, bump persistence.size before submitting. Growing a PVC after install works on most storage classes but is easier to get right up front.

Namespaces. The template picker defaults to your tenant's namespace. You can override that per-app if you want to isolate an app (e.g., "analytics" alongside "app-prod"), but only if that namespace already exists in the cluster — the console won't create one for you.

Duplicate installs. The picker disables cards for chart names you already have installed. If you actually want two copies of the same chart (e.g., a staging Postgres and a prod Postgres), use the custom chart page and give them different app names.

What's NOT in the template list

Templates are curated. They're not everything you can install. If you want Elasticsearch, ClickHouse, RabbitMQ, Vault, Jaeger, MongoDB, MySQL, MariaDB, Nextcloud, Rocket.Chat, Mattermost, Gitea, or any of the thousands of other public Helm charts — use the custom chart page. The provisioning code is identical; you just fill in the four fields the templates would have filled for you.