datus-dws

Huawei Cloud GaussDB(DWS) adapter for Datus.

DWS is a shared-nothing MPP analytical warehouse that speaks the PostgreSQL wire protocol. This adapter builds on datus-postgresql: it reuses the psycopg2 transport unchanged and overrides only what DWS does differently — its system schema set, its native table-definition function, and the Oracle-compatibility semantics of ORA mode.

Timeouts

timeout_seconds must be positive and now bounds driver I/O as well as pool checkout where applicable. Long operations may need a larger value. Server execution deadlines and supported overrides are described in the shared timeout policy and real-database validation. A socket/RPC timeout is distinct from an overall operation deadline.

Installation

pip install datus-dws

Configuration

datasources:
  dws_analytics:
    type: dws
    host: example.dws.myhuaweicloud.com
    port: 8000
    database: gaussdb
    schema: public
    username: dbadmin
    password: ${DWS_PASSWORD}
    sslmode: verify-ca
    sslrootcert: /path/to/cacert.pem
    timeout_seconds: 30

host also accepts the console's host:port form, in which case port may be omitted. The cluster's default database is normally gaussdb.

TLS

Two things to know about DWS certificates:

• verify-full is not usable. The default server certificate has CN=server and carries no subjectAltName, so hostname verification can never match a real endpoint. This is a property of the certificate, not a misconfiguration.
• Use the v2 CA. The console's dws_ssl_cert bundle contains both v1/sslcert/cacert.pem and v2/sslcert/cacert.pem. Only v2 matches the server certificate issuer; v1 is CN=Huawei Equipment CA and fails verification.

sslrootcert accepts either a filesystem path or inline PEM content, so a hosted deployment can pass an uploaded certificate directly.

Compatibility modes

DWS databases run in ORA, TD or MySQL compatibility mode, reported by pg_database.datcompatibility and surfaced through DWSConnector.get_compatibility_mode().

ORA mode — the default for new clusters — changes expression semantics in ways that can silently produce wrong results:

• 7/2 is 3.5 (double precision), not integer 3.
• '' is stored as NULL and '' IS NULL is true; col = '' never matches.
• 'a' || NULL is 'a'; concatenation absorbs NULL.
• DATE is timestamp(0) and DATE - DATE yields an interval.

The cloud integration suite runs the same contract and TPC-H tests in ORA, TD and MySQL modes.

Table DDL

Table DDL comes from DWS's pg_get_tabledef(), which preserves orientation, compression, DISTRIBUTE BY, partitioning, TABLESPACE and TO GROUP.

TABLESPACE and TO GROUP name objects of the source cluster and will not replay elsewhere. Use DWSConnector.strip_cluster_specific_clauses() to remove them before applying the DDL to a migration target.

Testing

# Unit tests (no cluster required)
python -m pytest tests/unit/ -v

# Integration tests against a live cluster
export DWS_HOST=... DWS_PORT=8000 DWS_DATABASE=gaussdb \
       DWS_USERNAME=dbadmin DWS_PASSWORD=... DWS_SSLROOTCERT=/path/to/cacert.pem
python -m pytest tests/integration/ -v

Integration tests skip automatically when DWS_HOST is unset. They create objects in a run-scoped schema and drop it on teardown.

Initialize persistent TPC-H sample data

The integration suite loads a deterministic five-table TPC-H fixture into its temporary schema. To initialize the same dataset in a schema kept for Agent testing, run from the datus-dws directory:

export DWS_HOST=... DWS_PORT=8000 DWS_DATABASE=gaussdb \
       DWS_USERNAME=dbadmin DWS_PASSWORD=... DWS_SCHEMA=main
python scripts/init_tpch_data.py --drop

The command reuses datus_dws.tpch_data, verifies every row count, and only drops the five tpch_* tables when --drop is explicitly provided.

Verified against

DWS 9.1.0.227, storage-decoupled (compute-group mode), 3 CN + 3 DN, ORA compatibility mode.