Release Notes 1.3

New Features

ADBC Netezza — -C adbc_netezza

FastBCP now supports Netezza / IBM NPS through the embedded ArpeNetezza ADBC driver, via the new -C adbc_netezza connection type. The driver is used for both the metadata step and the data step.

Key highlights:

• No NZdotNET installation needed
• Works with IBM Cloud NPS
• Parquet output only, like the other adbc_* connection types
• Supported parallel methods: None, NZDataSlice, Ntile, Random, RangeId and DataDriven
• Application Name defaults to FastBCP
• NPS 7.x: TLS 1.0/1.1 supported (tls_min_version=1.0, ArpeNetezza 0.3.2)

Connection string keys: --sourceconnectstring keys no longer need the adbc.arpenz. prefix. The following are all accepted:

• Friendly / ADO.NET keys: Server, Host, host:port, Port, Database, User ID, Password, SSL Mode, Application Name, …
• Plain driver keys: sslmode, tls_min_version, char_encoding, login_timeout, …
• uri and connection_string

Keys the driver does not support (e.g. propagate, Pooling) are ignored, with a single warning.

Windows

.\FastBCP.exe `
  --connectiontype adbc_netezza `
  --sourceconnectstring "Server=nps-host;Port=5480;Database=SALES;User ID=admin;Password=FastPassword" `
  --sourceschema "ADMIN" `
  --sourcetable "ORDERS" `
  --directory "D:\export\orders" `
  --fileoutput "orders.parquet" `
  --parallelmethod NZDataSlice `
  --paralleldegree 8

Linux

./FastBCP \
  --connectiontype adbc_netezza \
  --sourceconnectstring "Server=nps-host;Port=5480;Database=SALES;User ID=admin;Password=FastPassword" \
  --sourceschema "ADMIN" \
  --sourcetable "ORDERS" \
  --directory "/data/export/orders" \
  --fileoutput "orders.parquet" \
  --parallelmethod NZDataSlice \
  --paralleldegree 8

NPS 7 pre-releases

Files exported from NPS 7 with the pre-releases v1.3.0-netezza.1 / v1.3.0-netezza.2 may contain wrong values in NUMERIC columns. This is fixed in ArpeNetezza 0.3.3, shipped with FastBCP 1.3. Re-export the affected tables.

See Connection Parameters for more details.

New Parallel Method — DBPartition

The new -m DBPartition parallel method runs one extraction per native partition / subpartition of the source table.

Supported sources:

Source	Partitioning used
SQL Server	$PARTITION
PostgreSQL	Leaf tables
Oracle	PARTITION (...)
MySQL	PARTITION (...)
Teradata	Row partitioning, single- and multi-level PPI

The adbc_* counterparts are supported too.

• Requires --sourceschema and --sourcetable.
• Falls back to None, with a warning, when the table is not partitioned, has a single partition, or the source is not supported.
• The largest partitions are scheduled first (based on catalog row estimates).
• FASTBCP_DOP_EQUERY also applies to DBPartition.

The output layout is controlled by the new FASTBCP_DBPARTITION_TARGET_MODE environment variable:

Value	Output layout
CHUNK (default)	<file>_chunk_NNN.<ext>
HIVE	partition=<p>/[subpartition=<sp>/]<file>.<ext> (SQL Server uses bounds=<lower>_<upper>, Teradata L1=<n>/L2=<n>)

--merge works in both modes. See Parallel Parameters for more details.

Azure Blob Storage — Authentication and Container Access

For direct commit to Azure Blob Storage:

• Authentication: besides DefaultAzureCredential (still the fallback), FastBCP now accepts AZURE_STORAGE_CONNECTION_STRING or AZURE_STORAGE_SAS_URL. The selected authentication method is logged.
• Container access: a missing create permission on the container (HTTP 403/409) is tolerated when the container already exists. 403 errors now give a clearer message (firewall, RBAC or SAS permissions).

See Azure Configuration for more details.

ADBC Connectors Improvements

• --applicationintent with -C adbc_mssql: --applicationintent (default ReadOnly) and an ApplicationIntent key in --sourceconnectstring now reach the ArpeMSSQL driver. An Availability Group listener with read-only routing sends the export to a readable secondary, as with -C mssql.
• -C adbc_pgsql honours the libpq environment: PGUSER, PGPASSWORD, PGPASSFILE / ~/.pgpass and PGSSLROOTCERT are used when -U / -X / -g do not provide them, like -C pgsql (ArpePGSQL 0.4.5).
• -C adbc_oracle resolves TNS aliases through tnsnames.ora (TNS_ADMIN, $ORACLE_HOME/network/admin), like -C oraodp (ArpeOracle 0.3.11).

Behaviour Changes

adbc_mssql / adbc_pgsql / adbc_oracle Run Fully on ADBC

The metadata and planning step (row counts, column types, Ntile, Random, Timepartition, DBPartition, …) now runs on the ADBC driver instead of SqlClient, Npgsql or ODP.NET. A single connection string is used for both steps.

• --sourceconnectstring accepts the native syntax, explicit adbc.* / arpeio.* keys (they win over translated keys), or a whole-connection uri / adbc.<driver>.connection_string.
• Native keys with no driver equivalent (e.g. Pooling=false) are dropped with a warning. With a whole-connection uri / connection_string, native keys are dropped with a warning saying to put them inside the URI / connection string.
• A Driver= path in --sourceconnectstring is used by both steps.

Oracle Network Encryption with -C adbc_oracle

FastBCP now sends encryption=accepted and data_integrity=accepted, so the server decides, as with -C oraodp. FASTBCP_ADBC_ENCRYPTION / FASTBCP_ADBC_DATA_INTEGRITY or explicit keys still override this.

-C adbc_pgsql NUMERIC NaN

A NaN in a decimal column now fails the read instead of becoming NULL. Set adbc.arpepgsql.numeric_special=null in -g to keep the previous behaviour (ArpePGSQL 0.4.2+).

adbc_oracle DataDriven Keys

A scale-0 NUMBER(<=18) key is now treated as an integer, as with -C oraodp, so predicates and Hive paths are identical between both connection types.

NZDataSlice Scheduling

Slices now go through a shared work queue: an idle worker takes the next waiting slice, which is faster on uneven slices. After the first failure, no new slice starts. Output files are unchanged.

Bug Fixes

• Ntile lost rows (pre-existing): the last chunk now always includes its upper bound (rows were lost when the last bucket held a single key value), and a midnight lower bound no longer truncates the time of the upper bound. On Oracle, only date bounds are wrapped in TO_DATE (string keys broke before).
• Passwords in logs: quoted passwords containing ; (e.g. Password="p;w=1") are now fully masked in logs and telemetry, for every connection type.
• -C adbc_mssql with a named instance: host\INSTANCE connected to the instance on port 1433 instead of resolving the named instance through the SQL Server Browser.
• Oracle TNS alias with -U / -X: -C oraodp -S ALIAS -U u -X p failed with ORA-50008: 'USERNAME' is an invalid connection string attribute (same for adbc_oracle).
• -C nzoledb received a SQL Server connection instead of an OLE DB one.
• -C nzcopy -m Timepartition generated SQL Server syntax (YEAR() / WEEK()) instead of EXTRACT(...).
• NZDataSlice with --timestamped: all chunks of one export now share the same timestamp prefix (long exports were split across several prefixes).
• NZDataSlice log message fix.

Documentation

• AWS S3: IAM roles / policies needed by FastBCP and SSE-KMS support.
• FASTBCP_DBPARTITION_TARGET_MODE: new environment variable.
• Connection Parameters: adbc_netezza connection type.