Hosting the Lolly web app, handing out the packaged app and running a Lolly Work server are separate jobs. Start with the one you have, and read only the pages it needs. Choose a deployment compares the routes for the web app and the packaged apps.
Choose and install
- Install on Kubernetes: the Helm chart on Rancher, RKE2 or K3s, with your own PostgreSQL and identity provider.
- Run Lolly Work: the path that matches what you have, from Kubernetes to the public demo, with downloads and checksums.
- Install Lolly Work: from a source checkout to one server with Compose, systemd or YunoHost.
- Install the ready-made package: a finished Compose directory with HTTPS and password sign-in.
- How a Lolly Work deployment fits together: the parts every Lolly Work deployment runs and the ones you can leave out.
- Official installation routes: which packages exist, what each has passed and how to check a download.
Configure the instance
- Lolly Work instance configuration: every setting a Lolly Work instance reads.
- Configure your first Lolly Work instance: the first settings after install.
- Set up enterprise identity: single sign-on, SCIM and sign-out.
- Configure Lolly Work export signing: Content Credentials signed with your organisation's identity.
- Configure connector origins: which outside services the app may reach.
Keep it running
- Operations runbook: backups, restores, upgrades and recovery.
- Telemetry and dashboards: what the server measures and where you see the numbers.
- Metrics and alerts: Prometheus scraping, suggested alerts and a dashboard.
- Put a CDN in front of Lolly Work: which public files an edge may cache, and the checks before you turn one on.
- Audit log: the record of governed actions.
Build and package
- Build Guide: build the web app, the desktop and mobile apps and the terminal tools yourself.
- Build the web container and Helm chart: run Lolly on Kubernetes.
Warning: A recipe is not your server
An installation recipe describes a supported way to deploy. Before you update an instance that is already running, check the record of how that instance was deployed: its host, cluster, secrets, volumes and pinned release. A generic recipe cannot tell which server is your production one.