Connect Cdc Oracle

Redpanda Connect CDC: Oracle

The oracledb_cdc input in Redpanda Connect streams inserts, updates, and deletes from an Oracle database into Redpanda or any Kafka-compatible cluster using Oracle LogMiner. It can optionally snapshot all existing rows before streaming live changes. This is an Enterprise feature (Redpanda Community License) and requires Connect version 4.83.0 or later.

LogMiner reads Oracle's redo logs via the V$LOGMNR_CONTENTS view using the online_catalog strategy. The connector tracks progress with a System Change Number (SCN) stored either in a built-in Oracle checkpoint table (default) or in an external cache resource such as Redis or Memcached.

Quickstart

1. Prepare Oracle (four SQL commands)

-- 1. Verify or enable ARCHIVELOG mode (requires SYSDBA, then restart DB)
SELECT LOG_MODE FROM V$DATABASE;
-- If not ARCHIVELOG: SHUTDOWN IMMEDIATE; STARTUP MOUNT; ALTER DATABASE ARCHIVELOG; ALTER DATABASE OPEN;

-- 2. Enable minimal supplemental logging database-wide
ALTER DATABASE ADD SUPPLEMENTAL LOG DATA;

-- 3. Enable ALL columns supplemental logging for each table to capture
ALTER TABLE MYSCHEMA.ORDERS ADD SUPPLEMENTAL LOG DATA (ALL) COLUMNS;
ALTER TABLE MYSCHEMA.PRODUCTS ADD SUPPLEMENTAL LOG DATA (ALL) COLUMNS;

-- 4. Create a replication user and grant LogMiner privileges
CREATE USER rpcn IDENTIFIED BY "SecurePassword1";
GRANT CREATE SESSION TO rpcn;
GRANT SELECT ANY TRANSACTION TO rpcn;
GRANT LOGMINING TO rpcn;                     -- Oracle 12c+
GRANT SELECT ON V_$DATABASE TO rpcn;
GRANT SELECT ON V_$LOG TO rpcn;
GRANT SELECT ON V_$LOGFILE TO rpcn;
GRANT SELECT ON V_$ARCHIVED_LOG TO rpcn;
GRANT SELECT ON V_$ARCHIVE_DEST_STATUS TO rpcn;
GRANT SELECT ON V_$LOGMNR_CONTENTS TO rpcn;
GRANT SELECT ON ALL_TABLES TO rpcn;
GRANT SELECT ON ALL_LOG_GROUPS TO rpcn;
GRANT SELECT ON ALL_TAB_COLUMNS TO rpcn;
GRANT SELECT ON ALL_CONSTRAINTS TO rpcn;   -- snapshot PK discovery
GRANT SELECT ON ALL_CONS_COLUMNS TO rpcn;  -- snapshot PK discovery
GRANT SELECT ON MYSCHEMA.ORDERS TO rpcn;
GRANT SELECT ON MYSCHEMA.PRODUCTS TO rpcn;
-- For the built-in checkpoint table (default, no checkpoint_cache configured):
GRANT CREATE TABLE TO rpcn;
GRANT CREATE PROCEDURE TO rpcn;
-- Note: the CREATE USER statement above already creates the RPCN schema.
-- In Oracle, a schema is the same as a user and is created implicitly by CREATE USER.

2. Minimal pipeline YAML

# oracle-cdc.yaml
input:
  oracledb_cdc:
    connection_string: oracle://rpcn:SecurePassword1@oracle-host:1521/ORCL
    snapshot_mode: snapshot_and_stream   # snapshot existing rows, then stream (use snapshot_only for a one-time backfill)
    max_parallel_snapshot_tables: 2
    snapshot_max_batch_size: 1000
    include:
      - ^MYSCHEMA\.ORDERS$         # anchor with ^...$ to match exactly
      - ^MYSCHEMA\.PRODUCTS$
    logminer:
      scn_window_size: 20000       # default; tune up for high-volume
      backoff_interval: 5s
      mining_interval: 300ms
      strategy: online_catalog
      max_transaction_events: 0    # 0 = no limit
      lob_enabled: true
    checkpoint_limit: 1024

output:
  kafka_franz:
    seed_brokers:
      - redpanda-broker:9092
    topic: oracle-cdc-events

Run it:

rpk connect run oracle-cdc.yaml

3. Route each table to its own topic

input:
  oracledb_cdc:
    connection_string: oracle://rpcn:SecurePassword1@oracle-host:1521/ORCL
    include:
      - ^MYSCHEMA\.ORDERS$
      - ^MYSCHEMA\.PRODUCTS$
    logminer:
      scn_window_size: 20000
      backoff_interval: 5s
      mining_interval: 300ms
      strategy: online_catalog
      lob_enabled: true

pipeline:
  processors:
    - mapping: |
        meta topic = meta("table_name").lowercase()

output:
  kafka_franz:
    seed_brokers:
      - redpanda-broker:9092
    topic: ${! meta("topic") }

4. Use an external Redis checkpoint cache

cache_resources:
  - label: redis_scn_cache
    redis:
      url: redis://redis-host:6379

input:
  oracledb_cdc:
    connection_string: oracle://rpcn:SecurePassword1@oracle-host:1521/ORCL
    include:
      - ^MYSCHEMA\.ORDERS$
    logminer:
      scn_window_size: 20000
      backoff_interval: 5s
      mining_interval: 300ms
      strategy: online_catalog
      lob_enabled: true
    checkpoint_cache: redis_scn_cache
    checkpoint_cache_key: oracle-prod-scn
    checkpoint_limit: 1024

output:
  kafka_franz:
    seed_brokers:
      - redpanda-broker:9092
    topic: oracle-cdc-events

Message Metadata Fields

Every message emitted by oracledb_cdc carries these metadata fields (access with meta("field_name") in Bloblang):

Enterprise License

oracledb_cdc is an enterprise connector licensed under the Redpanda Community License. You need a valid Redpanda Enterprise license. Without it the connector refuses to start with a license error ("all enterprise connectors are blocked"). Set your license via the environment variable or contact Redpanda for a 30-day trial key.

Several Redpanda Enterprise features apply to a CDC pipeline and to the destination cluster where the stream lands. Each requires a valid Enterprise license:

• Iceberg Topics — materialize CDC changes as Apache Iceberg tables. Enable with iceberg_enabled (cluster) and redpanda.iceberg.mode (topic). Use the schema metadata field with schema_registry_encode for structured (value_schema_id_prefix) tables.
• Server-side Schema ID Validation — have brokers reject records with unregistered schema IDs. Enable with enable_schema_id_validation (cluster) and redpanda.value.schema.id.validation (topic).
• Tiered Storage — retain CDC history in object storage. Enable with cloud_storage_enabled (cluster) and redpanda.remote.write / redpanda.remote.read (topic).
• Connect enterprise capabilities — secrets management (for the Oracle password / wallet_password), the redpanda{} config service block (logs/status to a topic), allow/deny lists, and FIPS-compliant rpk connect.

See enterprise-features.md for every nested config key, mode value, and license-expiry behavior.

Reference Directory

• config-reference.md: Complete field reference for every oracledb_cdc config option, grounded in source — types, defaults, constraints, and the nested logminer{} sub-block.
• setup-oracle.md: Preparing Oracle: ARCHIVELOG mode, supplemental logging, LogMiner grants, Oracle Wallet for SSL, and CDB/PDB (pluggable database) notes.
• pipeline-and-output.md: Full runnable pipelines, message/metadata shape, per-table topic routing, LOB handling, snapshot-then-stream behavior, checkpointing, and restart/resume semantics.
• enterprise-features.md: Redpanda Enterprise features relevant to a CDC pipeline and their nested config keys — Iceberg Topics (redpanda.iceberg.mode/target.lag.ms/partition.spec/invalid.record.action, iceberg_enabled), server-side Schema ID Validation (enable_schema_id_validation, redpanda.{key,value}.schema.id.validation), Tiered Storage (cloud_storage_enabled, redpanda.remote.write/read), and Connect enterprise capabilities (secrets, the redpanda{} config service block, allow/deny lists, FIPS). Includes license-expiry behavior for each.