https://luth.io/tags/cdc/ Recent content in Cdc on Luther Monson Hugo en Tue, 15 Sep 2026 21:08:00 -0700 https://luth.io/blog/2026/09/clustered-sqlite-cdc-cluster-channel/ Tue, 15 Sep 2026 21:08:00 -0700 https://luth.io/blog/2026/09/clustered-sqlite-cdc-cluster-channel/ <p>By the end of <a href="https://luth.io/blog/2026/08/clustered-sqlite-adding-turso/">Part 2</a> ePHPm v0.7.0 ran exactly one SQLite engine in process. That was Turso, for single-node and clustered mode alike. The sidecar was gone. What I skipped over is how a replica actually gets the primary&rsquo;s writes, and what happens when the primary dies with data on it. That&rsquo;s this post.</p> <ul> <li><a href="https://luth.io/blog/2026/08/clustered-sqlite-sqld-sidecar/">Part 1: Replication by sidecar with sqld</a></li> <li><a href="https://luth.io/blog/2026/08/clustered-sqlite-adding-turso/">Part 2: Adding Turso</a></li> <li>Part 3: CDC over the cluster channel (this post)</li> </ul> <h3 id="cdc-from-the-engine-itself">CDC from the engine itself</h3> <p>Turso has change data capture built in. Turn it on for a connection and every committed change lands in a <code>turso_cdc</code> table. Each row has a monotonically increasing <code>change_id</code>, the table, the rowid, and before/after images of the record. The primary tails that table and ships transaction batches. Replicas apply them. That&rsquo;s the whole design. It landed as an experimental path in <a href="https://github.com/ephpm/ephpm/pull/186">#186</a> at the end of July. The tailer and applier live in litewire (<a href="https://github.com/ephpm/litewire/pull/10">litewire#10</a>).</p>