gRPC

GreptimeDB offers gRPC SDKs for efficient and high-performance data ingestion.

If there is no SDK available for your programming language, you have the option to create your own SDK by following the guidelines provided in the contributor guide.

Hints

Use the x-greptime-hints gRPC metadata to pass hints as a comma-separated list of key=value pairs.

Disable WAL for a single write request

Set insert_skip_wal=true in the x-greptime-hints gRPC metadata to disable the Write-Ahead Log (WAL) for the current insert request. For Bulk Write using Arrow Flight DoPut, this hint applies to the entire stream.

This setting applies only to the current request and does not change the table-level skip_wal option. Setting insert_skip_wal to false or omitting the hint does not enable WAL if the table-level skip_wal option is true.

warning

When WAL is disabled, unflushed data is lost if the process restarts. Use this option only when the data can be ingested again from its source.

Bulk Write schema updates

Arrow Flight DoPut Bulk Write requires an existing target table. By default, the server checks the stream's Arrow schema before processing the first data batch and adds missing columns to the table. The frontend/standalone top-level auto_create_table configuration and request metadata x-greptime-hint-auto_create_table both default to true. Setting either to false disables automatic schema updates for ordinary tables. A request value of true does not override a server value of false. New columns are fields, not primary-key columns. Define primary-key columns with SQL before writing.

Automatic column addition does not support nested Arrow types or Dictionary types. Input must satisfy the target table's type, time-index, and nullability constraints.