Choose the permanent identity
Plan the domain, registration policy, community rules and transactional email before inviting people. Make recovery possible without relying entirely on the instance itself.
Size the working set
The Guardian tier is a starting catalogue candidate for evaluation. User count alone does not predict load: federation, media, database growth and background processing can dominate.
Keep the stack maintainable
Use supported application, database and cache versions. Secure the reverse proxy, protect administrator accounts and monitor queues and disk capacity.
Moderation is part of hosting
Set a reachable contact channel and a consistent process for reports and federation decisions. Minimal hosting signup does not remove the instance operator’s duties to the community.
Recover the complete instance
Keep consistent copies of database, media, configuration and signing secrets. Test restoration into an isolated environment before a real outage requires it.
Choose resources for this workload
Guardian is a reasonable catalogue point to evaluate for an application database, workers and local media together. A real community needs measurements of active users, remote media, queue depth and retention; do not translate a RAM tier into a guaranteed user count.
A community instance rollout
Use the maintained upstream instructions for the selected release. The sequence below identifies the decisions and pass conditions to record in your own runbook.
- Choose the public domain and registration policy before installation
- Prepare supported application dependencies, PostgreSQL and Redis
- Configure HTTPS, external mail delivery and protected secrets
- Create the administrator account and complete the instance settings
- Test a real message, media upload, background job and federated interaction
- Back up database, required secrets and media according to the recovery plan
Official documentation: Maintained upstream setup guidance
What to verify before relying on it
Plan moderation and account support as part of operations. Add object storage or more workers because measurements require it, and test recovery before the instance becomes someone else’s only account.
A little more clarity.
Does this use case include a managed application installation?
No. It describes a workload for customer-administered infrastructure. Application setup, updates, moderation, licensing and recovery remain your responsibility unless a separate managed agreement says otherwise.