ingestr

Appearance

IBM Db2 ​

IBM Db2 is a high-performance, enterprise-grade relational database system designed for reliability, scalability, and transactional integrity.

ingestr supports IBM Db2 as a source.

Installation ​

Db2 source support is built into the ingestr binary. It does not require the IBM CLI driver, CGO, or a separate Python add-on.

URI format ​

The URI format for DB2 is as follows:

plaintext
db2://<username>:<password>@<host>:<port>/<database-name>

URI parameters:

  • username: The username to connect to the database
  • password: The password for the user
  • host: The host address of the database server
  • port: The port number the database server is listening
  • database-name: the name of the database to connect to

Optional query parameters:

  • schema: default schema to use for unqualified table names. If omitted, ingestr uses the username uppercased.
  • ssl: set to true or 1 to connect with TLS.
  • timeout: connection/read timeout in seconds.
  • platform: forces which Db2 catalog dialect is used when reading table metadata. One of luw, zos, or ibmi. ingestr auto-detects this from the server's PRDID during the connection handshake, so this is only needed if detection picks the wrong dialect.
  • rdb_name (Db2 for i only): overrides the real RDB directory name used for internal package lookups. ingestr auto-detects this from the server's handshake response, so this is normally not needed - see note below.

Platform support ​

ingestr's Db2 client speaks DRDA directly, the wire protocol shared by every Db2 platform, so the same db2:// URI works against:

  • Db2 for Linux/Unix/Windows (LUW)
  • Db2 for z/OS
  • Db2 for i (iSeries/AS400)

The platform is detected automatically from the server's product ID during the handshake and determines which system catalog (SYSCAT.* for LUW/z/OS, QSYS2.* for i) is used to read column and primary key metadata.

For Db2 for i, database is commonly a symbolic alias such as *LOCAL (resolved by the server during the connection itself), but ingestr's internal package lookups need the real RDB directory name instead (the value shown by WRKRDBDIRE on the host). ingestr auto-detects this from the handshake response, but if that ever fails, set it explicitly with rdb_name, e.g. db2://user:pass@host:446/%2ALOCAL?platform=ibmi&rdb_name=RDBNAME2.

Status and known limitations ​

The DRDA/DDM wire protocol implementation and the Db2 for i (iSeries/AS400) catalog/handshake path have been exercised against a live production Db2 for i server, including CHAR/VARCHAR/VARGRAPHIC, DATE/TIME/TIMESTAMP, INTEGER/SMALLINT/BIGINT/FLOAT, packed and zoned DECIMAL/NUMERIC, multi-block result sets, and SECMEC 8 (password-substitute) authentication. Db2 for Linux/Unix/Windows and Db2 for z/OS use the same DRDA client and the same catalog-dialect switch, but that code path has only been exercised against a fake DRDA server in unit tests, not a live LUW or z/OS server - if you hit something, please open an issue with the platform and error.

Not yet supported:

  • CLOB, DBCLOB, and BLOB columns delivered via an SDA (Self-Defining type) descriptor. These fail with a clear error naming the column rather than silently corrupting the row; exclude the column from the query as a workaround.

Known gaps in test coverage (works as implemented, but not independently confirmed against a live server capture):

  • The VARGRAPHIC catalog type name truncation (VARG) is handled on the same evidence pattern as a confirmed case (TIMESTMP), but hasn't itself been confirmed against a fresh catalog query.
  • SQLSTATE in the SQLCARD diagnostic structure is decoded as CCSID 500, consistent with every other DRDA protocol-metadata field, but hasn't been confirmed against a real non-ASCII SQLSTATE value.