Skip to main content

Enable CDF

Step 4 — Enable Lakebase CDF on lakets_cdf​

Step 3.6 (enable_sync) created the unpartitioned shadow in the lakets_cdf schema — with REPLICA IDENTITY FULL set automatically — and its mirror trigger. LakeTS cannot turn on this managed sync itself, so enable it once in the Databricks UI:

  1. Open the Lakebase project, then Branch overview → CDF.
  2. Click Start sync and select the lakets_cdf schema.
  3. Set the destination Unity Catalog catalog and schema.
  4. Start the sync.

CDF snapshots the shadow, then switches to continuous streaming. Without this step the shadow fills but never reaches Unity Catalog, so no chunk is ever flagged or dropped — the durability gate is never satisfied. Confirm it is streaming:

SELECT table_oid::regclass AS shadow, status, committed_lsn
FROM wal2delta.tables;   -- status: SNAPSHOTTING, then STREAMING

See Lakebase CDF setup for the full walkthrough.