Using CSV Log Output

Prerequisites

• The log_destination parameter is set to csvlog.
• The logging_collector parameter is set to on.

Definition of csvlog

Log lines are emitted in comma separated values (CSV) format.

An example table definition for storing CSV-format log output is shown as follows:

CREATE TABLE gaussdb_log
(
log_time timestamp(3) with time zone,
node_name text,
user_name text,
database_name text,
process_id bigint,
connection_from text,
"session_id" text,
session_line_num bigint,
command_tag text,
session_start_time timestamp with time zone,
virtual_transaction_id text,
transaction_id bigint,
query_id bigint,
module text,
error_severity text,
sql_state_code text,
message text,
detail text,
hint text,
internal_query text,
internal_query_pos integer,
context text,
query text,
query_pos integer,
location text,
application_name text
);

For details, see Table 1.

Table 1 Meaning of each csvlog field

Run the following command to import a log file to this table:

COPY gaussdb_log FROM '/opt/data/pg_log/logfile.csv' WITH csv;

NOTE: The log name (logfile.csv) here needs to be replaced with the name of a log generated.

Simplifying Input

Simplify importing CSV log files by performing the following operations:

• Set log_filename and log_rotation_age to provide a consistent, predictable naming solution for log files. By doing this, you can predict when an individual log file is complete and ready to be imported.
• Set log_rotation_size to 0 to disable size-based log rollback, as it makes the log file name difficult to predict.
• Set log_truncate_on_rotation to on so that old log data cannot be mixed with the new one in the same file.