ob-jobs/ob_jobs/query_log.py

from ob_jobs.common import ChOptions


def query_log_source_table(ch_opts: ChOptions):
    tmp_table_name = 't_query_log'
    sql = f"""
    create table {tmp_table_name} (
    `hostname` STRING,
    `type` STRING,
    `event_date` DATE,
    `event_time` TIMESTAMP,
    `query_start_time` TIMESTAMP,
    `query_duration_ms` DECIMAL,
    `read_rows` DECIMAL,
    `read_bytes` DECIMAL,
    `written_rows` DECIMAL,
    `written_bytes` DECIMAL,
    `result_rows` DECIMAL,
    `result_bytes` DECIMAL,
    `memory_usage` DECIMAL,
    `current_database` STRING,
    `query` STRING,
    `formatted_query` STRING,
    `normalized_query_hash` DECIMAL(20, 0),
    `query_kind` STRING,
    `databases` ARRAY<STRING>,
    `tables` ARRAY<STRING>,
    `columns` ARRAY<STRING>,
    `partitions` ARRAY<STRING>,
    `exception_code` INTEGER,
    `exception` STRING,
    `stack_trace` STRING,
    `is_initial_query` SMALLINT,
    `user` STRING,
    `query_id` STRING,
    `port` INTEGER,
    `client_name` STRING,
    `http_user_agent` STRING
    ) with (
       'connector' = 'clickhouse',
       'url' = 'clickhouse://{ch_opts.ch_host}:{ch_opts.ch_port}',
       'database-name' = '{ch_opts.ch_dbname}',
       'table-name' = 'query_log',
       'username' = '{ch_opts.ch_user}',
       'password' = '{ch_opts.ch_password}'
   )
    """
    return sql, tmp_table_name