Tailscale identified a SQLite write‑ahead logging race condition that caused repeated database corruption in its control plane over a six‑month period from late 2023 into 2024. The issue affected the shards that host internal SQLite databases used for tailnet metadata. Corruption led to control‑plane outages lasting up to an hour, preventing new devices from joining the network and temporarily blocking access to the admin console and API. Tailscale’s engineering team, together with the SQLite developers, traced the problem to a data race between a checkpoint operation and a write transaction. The race occurred only when manual checkpointing was performed at a high frequency. After extensive telemetry collection and analysis, the SQLite developers released version 3.51.3, which adds a check to detect WAL resets during checkpointing. Tailscale deployed the fix across its shards, reduced timestamp precision to avoid a secondary rounding issue, and added warning logs to detect overlapping write and checkpoint operations. Since the fix was applied, no further corruption incidents have been observed for four months. The company noted that the incident highlighted the risk of using non‑standard database configurations and reinforced the importance of standard operational paths. Future improvements include enhanced backup verification and continued collaboration with the SQLite community.