Tailscale has attributed a series of outages and database-corruption incidents beginning in August 2025 to a rare flaw in SQLite’s write-ahead logging system that had apparently remained undiscovered for 16 years. The issue matters because SQLite stored information for Tailscale’s private mesh networks, so corruption could disrupt its control-plane services and affect users’ connectivity.

Tailscale had used SQLite as its primary database since 2022 and took full database snapshots every few minutes before uploading them to Amazon S3. After six months of investigation with SQLite maintainers, the teams identified a data race between database checkpointing and a write transaction, known as the “WAL-Reset bug”. They created a specialised virtual file-system logging tool, funded by Tailscale, to capture the rare event after earlier theories involving file locks, memory and thread safety were ruled out.