Getting started · How it works

Getting started

How it works

BackupLatte reads supported Linear data through Linear's API, stores readable snapshots, and can restore supported parts into Linear when you ask it to.

A backup reads the Linear workspace and stores a snapshot; a snapshot can be downloaded or restored into Linear

Connection

You connect a workspace once, through Linear's own authorisation screen (OAuth). BackupLatte receives an access token for that workspace — never your Linear password — and stores it encrypted. Everything it does in Linear it does as the Linear user who connected it, so a backup collects supported data visible to that user, subject to permissions and API availability. See Connect a workspace.

A backup run

  1. Queued. At the top of every hour BackupLatte queues the workspaces whose daily hour has come. A workspace that already has a backup waiting or running is skipped, and so is one whose owner has no active plan. Run backup now queues a backup straight away.
  2. Collected. The worker reads the supported sets of the workspace — issues, comments, projects, documents, teams, cycles and the rest — archived records included. Linear limits how fast apps may read; when Linear asks it to slow down, the worker waits and carries on.
  3. Files copied. Images and files uploaded to Linear that the content links to are copied too. Each file is stored once per workspace and shared by every snapshot that links to it.
  4. Checked. Linear has no point-in-time snapshot, and people keep working while a backup runs. At the end, records changed during the run are read again, where Linear allows filtering by change time. This narrows the differences between sets, but a backup is not an atomic point-in-time copy of the whole workspace.
  5. Stored. Each set is saved as gzipped JSON, next to a manifest that records, per set, its state, record count, size and SHA-256 checksum.

The backup ends as Success when the required sets are complete or unavailable on the provider plan, Partial (with gaps) when a required set could not be collected in full, or Failed when the run cannot finish successfully. Collected data in a readable partial snapshot can be downloaded; only supported, complete datasets can be selected for restore. See Snapshots and downloads.

Self-contained snapshots

A readable snapshot contains the collected data rather than only changes since the previous run. It can be downloaded independently, but a Partial snapshot has recorded gaps and restores are limited to supported, complete datasets. Check its contents and manifest. Files are shared: a copy already stored for the workspace is reused rather than stored again for every snapshot.

Cleanup

Once a day, at 03:00 UTC, snapshots older than the workspace's retention period are deleted — see Retention. Unreferenced stored files become eligible for background cleanup.

Restore

A restore reads a snapshot and creates its records in a Linear workspace you have connected — a new, empty one or the one it came from — in dependency order, from teams to comments. It saves its progress as it goes. A resumable failed restore reuses saved mappings and target-scoped record IDs rather than deliberately creating those records again. See Restore.

Where it runs

The application, its database and background jobs run on servers in Germany (Hetzner, Falkenstein). Snapshots and copied files are stored in Cloudflare R2 under its EU jurisdiction. More in Security.

ClickUp

ClickUp uses its own authorization and dataset coverage. Its scheduler checks at minute 10 of each hour and retention cleanup runs at 03:10 UTC. Recovery selects lists and Docs and creates new records; Linear's matching and duplicate-avoidance rules do not apply. Start with ClickUp contents and restore.

Something missing or unclear? Write to [email protected].