文章介绍了四种将CSV格式的PostgreSQL日志导入数据库的方法，包括使用file_fdw扩展、csvsql工具、awk命令以及手动创建表结构和COPY命令。每种方法都有其优缺点，如file_fdw限制于单个文件，csvsql可能导致数据类型错误，而awk则能生成SQL脚本进行批量导入。最佳实践是手动创建表结构并用COPY命令导入数据。
将pg日志导入pg数据库的几种方法
注意：我这里导入的pg日志是csv格式的，Postgresql 14.x日志的字段是26个，网上很多文章都是23个字段
1 方法1：官方方法（fiel_fdw）
参考：https://www.postgresql.org/docs/14/file-fdw.html
结论：好像只能一次一个文件，如何将多个日志文件导入到一个数据库中？out！
1.1 创建file_fdw数据源（在此之前需要安装pg的file_fdw扩展）
CREATE SERVER pg_csvlog_fdw FOREIGN DATA WRAPPER file_fdw;
1.2 创建外部表并导入数据
CREATE FOREIGN TABLE pg_csvlog_file_fdw (
log_time timestamp(3) with time zone,
user_name text,
database_name text,
process_id integer,
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,
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,
backend_type text,
leader_pid integer,
query_id bigint
) SERVER pg_csvlog_fdw
OPTIONS ( filename '/usr/local/pgsql/data/log/postgresql-2023-04-21_130000.csv', format 'csv' );
# 你看，不能copy，有其他方法吗？我不会
pg_log=# copy "pg_csvlog_file_fdw" from '/usr/local/pgsql/data/log/postgresql-2023-04-21_160000.csv' with csv;
ERROR: cannot insert into foreign table "pg_csvlog_file_fdw"
- 查看刚创建的表结构：
pg_log=# \d pg_csvlog_file_fdw
Foreign table "public.pg_csvlog_file_fdw"
Column | Type | Collation | Nullable | Default | FDW options
------------------------+-----------------------------+-----------+----------+---------+-------------
log_time | timestamp(3) with time zone | | | |
user_name | text | | | |
database_name | text | | | |
process_id | integer | | | |
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 | | | |
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 | | | |
backend_type | text | | | |
leader_pid | integer | | | |
query_id | bigint | | | |
Server: pg_csvlog_fdw
FDW options: (filename '/usr/local/pgsql/data/log/postgresql-2023-04-21_130000.csv', format 'csv')
2 方法2：用csvsql工具导入
参考：csvkit - 处理csv文件的瑞士军刀： https://segmentfault.com/a/1190000009179225
结论：用它创建的表，字段值可能被重新定义，比如transaction_id的值等于0（他会分析多条记录得出），他会把它当成boolean型，并在导入时直接把插入值改为false，与原值0就会不匹配，如果手动校正数据类型为bigint型，也不行，因为导入数据时，他依然会将插入的值改为false，此时导入会出错。
2.1 pg csv log 表头准备
log_time,user_name,database_name,process_id,connection_from,session_id,session_line_num,command_tag,session_start_time,virtual_transaction_id,transaction_id,error_severity,sql_state_code,message,detail,hint,internal_query,internal_query_pos,context,query,query_pos,location,application_name,backend_type,leader_pid,query_id
2.2 为 postgresql-2023-04-21_140000.csv 添加表头行(26列)
sed -i '1ilog_time,user_name,database_name,process_id,connection_from,session_id,session_line_num,command_tag,session_start_time,virtual_transaction_id,transaction_id,error_severity,sql_state_code,message,detail,hint,internal_query,internal_query_pos,context,query,query_pos,location,application_name,backend_type,leader_pid,query_id
' postgresql-2023-04-21_140000.csv
2.3 根据pg日志文件，输出创建表的SQL语句（用postgresql数据库格式），并用它创建表
^_^ kevin@TM1701-b38cbc23:~$ csvsql -i postgresql --tables pg_csvlog_csvsql /usr/local/pgsql/data/log/postgresql-2023-04-21_140000.csv
CREATE TABLE pg_csvlog (
log_time VARCHAR NOT NULL,
user_name VARCHAR,
database_name VARCHAR,
process_id DECIMAL NOT NULL,
connection_from VARCHAR,
session_id VARCHAR NOT NULL,
session_line_num DECIMAL NOT NULL,
command_tag VARCHAR,
session_start_time VARCHAR NOT NULL,
virtual_transaction_id VARCHAR,
transaction_id BOOLEAN NOT NULL,
error_severity VARCHAR NOT NULL,
sql_state_code VARCHAR NOT NULL,
message VARCHAR NOT NULL,
detail VARCHAR,
hint BOOLEAN,
internal_query BOOLEAN,
internal_query_pos BOOLEAN,
context BOOLEAN,
query VARCHAR,
query_pos BOOLEAN,
location BOOLEAN,
application_name VARCHAR,
backend_type VARCHAR NOT NULL,
leader_pid BOOLEAN,
query_id BOOLEAN NOT NULL
);
这里有个问题，他把【transaction_id】弄成了boolean型，在下面导入时也转换成boolean型了(true或false，不是原本的0或1)，所以修改表结构也没用，讨厌，他本来是int型的，这个与sqlalchemy有关，我又不会弄
2.4 导入数据
csvsql --db postgresql://postgres:POSTGRES@192.168.11.145:5432/pg_log --tables pg_csvlog_csvsql --insert /usr/local/pgsql/data/log/postgresql-2023-04-21_140000.csv --no-create
可以导入多个日志文件，但每个文件都要重复执行添加表头命令
3 方法3：用awk方法导入
结论：这个方法也是OK的。创建表，用awk生成sql语法文件，然后导入
3.1 根据官方文档创建数据表
CREATE TABLE pg_csvlog_awk (
log_time timestamp(3) with time zone,
user_name text,
database_name text,
process_id integer,
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,
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,
backend_type text,
leader_pid integer,
query_id bigint,
PRIMARY KEY (session_id, session_line_num)
);
3.2 用csv日志文件生成SQL语法文件
awk -F, '{ print "insert into pg_csvlog_awk ( log_time, user_name, database_name, process_id, connection_from, session_id, session_line_num, command_tag, session_start_time, virtual_transaction_id, transaction_id, error_severity, sql_state_code, message, detail, hint, internal_query, internal_query_pos, context, query, query_pos, location, application_name, backend_type, leader_pid, query_id ) values" , "(" "'\''"$1"'\''" "," "'\''"$2"'\''" "," "'\''"$3"'\''" "," "'\''"$4"'\''" "," "'\''"$5"'\''" "," "'\''"$6"'\''" "," "'\''"$7"'\''" "," "'\''"$8"'\''" "," "'\''"$9"'\''" "," "'\''"$10"'\''" "," "'\''"$11"'\''" "," "'\''"$12"'\''" "," "'\''"$13"'\''" "," "'\''"$14"'\''" "," "'\''"$15"'\''" "," "'\''"$16"'\''" "," "'\''"$17"'\''" "," "'\''"$18"'\''" "," "'\''"$19"'\''" "," "'\''"$20"'\''" "," "'\''"$21"'\''" "," "'\''"$22"'\''" "," "'\''"$23"'\''" "," "'\''"$24"'\''" "," "'\''"$25"'\''" "," "'\''"$26"'\''" ")" ";" }' /usr/local/pgsql/data/log/postgresql-2023-04-21_140000.csv > pg_csvlog_awk.sql
用\换行容易理解：
awk -F , '{ \
print "insert into pg_csvlog_awk ( \
log_time, user_name, database_name, process_id, connection_from, session_id, session_line_num, command_tag, session_start_time, virtual_transaction_id, transaction_id, error_severity, sql_state_code, message, detail, hint, internal_query, internal_query_pos, context, query, query_pos, location, application_name, backend_type, leader_pid, query_id \
) values" , "(" \
"'\''"$1"'\''" "," "'\''"$2"'\''" "," "'\''"$3"'\''" "," "'\''"$4"'\''" "," "'\''"$5"'\''" "," "'\''"$6"'\''" "," "'\''"$7"'\''" "," "'\''"$8"'\''" "," "'\''"$9"'\''" "," "'\''"$10"'\''" "," "'\''"$11"'\''" "," "'\''"$12"'\''" "," "'\''"$13"'\''" "," "'\''"$14"'\''" "," "'\''"$15"'\''" "," "'\''"$16"'\''" "," "'\''"$17"'\''" "," "'\''"$18"'\''" "," "'\''"$19"'\''" "," "'\''"$20"'\''" "," "'\''"$21"'\''" "," "'\''"$22"'\''" "," "'\''"$23"'\''" "," "'\''"$24"'\''" "," "'\''"$25"'\''" "," "'\''"$26"'\''" \
")" ";" \
}' /usr/local/pgsql/data/log/postgresql-2023-04-21_140000.csv > pg_csvlog_awk.sql
3.3 导入SQL文件
psql pg_csvlog_awk < pg_csvlog_awk.sql
4 方法4：最佳实践
根据官方文档手动创建表结构，然后用copy方式导入日志文件
4.1 创建表
CREATE TABLE IF NOT EXISTS pg_csvlog_shell (
log_time timestamp(3) with time zone,
user_name text,
database_name text,
process_id integer,
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,
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,
backend_type text,
leader_pid integer,
query_id bigint,
PRIMARY KEY (session_id, session_line_num)
);
4.2 添加字段说明（多此一举）
comment on column pg_csvlog_shell.log_time is '日志时间，带毫秒的时间戳';
comment on column pg_csvlog_shell.user_name is '当前登录数据库的用户名';
comment on column pg_csvlog_shell.database_name is '数据库名';
comment on column pg_csvlog_shell.process_id is '进程ID';
comment on column pg_csvlog_shell.connection_from is '客户端主机:端口号';
comment on column pg_csvlog_shell.session_id is '会话ID 由后台进程启动时间和PID组成';
comment on column pg_csvlog_shell.session_line_num is '每个会话的行号，类似history命令';
comment on column pg_csvlog_shell.command_tag is '命令标签';
comment on column pg_csvlog_shell.session_start_time is '会话开始时间';
comment on column pg_csvlog_shell.virtual_transaction_id is '虚拟事务ID';
comment on column pg_csvlog_shell.transaction_id is '事务ID';
comment on column pg_csvlog_shell.error_severity is '错误等级';
comment on column pg_csvlog_shell.sql_state_code is 'SQLSTATE 代码';
comment on column pg_csvlog_shell.message is '消息 SQL语句';
comment on column pg_csvlog_shell.detail is '错误消息详情';
comment on column pg_csvlog_shell.hint is '提示';
comment on column pg_csvlog_shell.internal_query is '导致错误的内部查询（如果有）';
comment on column pg_csvlog_shell.internal_query_pos is '错误位置所在的字符计数';
comment on column pg_csvlog_shell.context is '错误上下文';
comment on column pg_csvlog_shell.query is '导致错误的用户查询（如果有且被log_min_error_statement启用）';
comment on column pg_csvlog_shell.query_pos is '错误位置所在的字符计数';
comment on column pg_csvlog_shell.location is '在 PostgreSQL 源代码中错误的位置（如果log_error_verbosity被设置为verbose）';
comment on column pg_csvlog_shell.application_name is '应用名';
comment on column pg_csvlog_shell.backend_type is '后端类型（新加）';
comment on column pg_csvlog_shell.leader_pid is 'Leader进程ID（新加）';
comment on column pg_csvlog_shell.query_id is '查询ID（新加）';
4.3 导入数据
pg_log=# copy "pg_csvlog_shell" from '/usr/local/pgsql/data/log/postgresql-2023-04-21_160000.csv' with csv;
COPY 1165
4.4 直接给个脚本吧
[postgres@v-192-168-11-145-pg pg]$ cat import_pg_log.sh
#!/bin/bash
echo
echo "用法： $0 <csv日志清单文件> #-- 默认导入今天所有的日志"
echo
LOG_PATH="/usr/local/pgsql/data/log"
LOG_FILE_LIST="pg_log.list"
LOG_ANALYSE_DB="pg_ana_log"
LOG_ANALYSE_TABLE="pg_csvlog_shell"
TIME=${TIME:-`date +%Y-%m-%dT%H:%M:%S`}
TIME_START=${TIME}
DATE_DAY=`date -d "${TIME}" +%Y%m%d`
LOG_ANALYSE_TABLE="${LOG_ANALYSE_TABLE}_${DATE_DAY}"
if [[ $# == 0 ]]; then
ls -l ${LOG_PATH}/postgresql-$(date -d "${TIME}" +%Y-%m-%d)_*csv | awk '{print $9}' > ${LOG_FILE_LIST}
else
LOG_FILE_LIST=$1
fi
echo ""
echo "##################################################"
echo "说明："
echo "数据库名：${LOG_ANALYSE_DB}"
echo "数据库名：${LOG_ANALYSE_TABLE}"
echo "数据库日志文件路径：${LOG_PATH}"
echo "数据库日志文件清单："
cat ${LOG_FILE_LIST}
echo ""
read -t 10 -p '按任意键继续......'
echo "##################################################"
echo "创建库表"
/usr/local/pgsql/bin/psql << EOF
DROP DATABASE "${LOG_ANALYSE_DB}";
CREATE DATABASE "${LOG_ANALYSE_DB}";
\c "${LOG_ANALYSE_DB}"
CREATE TABLE IF NOT EXISTS "${LOG_ANALYSE_TABLE}" (
log_time timestamp(3) with time zone,
user_name text,
database_name text,
process_id integer,
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,
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,
backend_type text,
leader_pid integer,
query_id bigint,
PRIMARY KEY (session_id, session_line_num)
);
comment on column ${LOG_ANALYSE_TABLE}.log_time is '日志时间，带毫秒的时间戳';
comment on column ${LOG_ANALYSE_TABLE}.user_name is '当前登录数据库的用户名';
comment on column ${LOG_ANALYSE_TABLE}.database_name is '数据库名';
comment on column ${LOG_ANALYSE_TABLE}.process_id is '进程ID';
comment on column ${LOG_ANALYSE_TABLE}.connection_from is '客户端主机:端口号';
comment on column ${LOG_ANALYSE_TABLE}.session_id is '会话ID 由后台进程启动时间和PID组成';
comment on column ${LOG_ANALYSE_TABLE}.session_line_num is '每个会话的行号，类似history命令';
comment on column ${LOG_ANALYSE_TABLE}.command_tag is '命令标签';
comment on column ${LOG_ANALYSE_TABLE}.session_start_time is '会话开始时间';
comment on column ${LOG_ANALYSE_TABLE}.virtual_transaction_id is '虚拟事务ID';
comment on column ${LOG_ANALYSE_TABLE}.transaction_id is '事务ID';
comment on column ${LOG_ANALYSE_TABLE}.error_severity is '错误等级';
comment on column ${LOG_ANALYSE_TABLE}.sql_state_code is 'SQLSTATE 代码';
comment on column ${LOG_ANALYSE_TABLE}.message is '消息 SQL语句';
comment on column ${LOG_ANALYSE_TABLE}.detail is '错误消息详情';
comment on column ${LOG_ANALYSE_TABLE}.hint is '提示';
comment on column ${LOG_ANALYSE_TABLE}.internal_query is '导致错误的内部查询（如果有）';
comment on column ${LOG_ANALYSE_TABLE}.internal_query_pos is '错误位置所在的字符计数';
comment on column ${LOG_ANALYSE_TABLE}.context is '错误上下文';
comment on column ${LOG_ANALYSE_TABLE}.query is '导致错误的用户查询（如果有且被log_min_error_statement启用）';
comment on column ${LOG_ANALYSE_TABLE}.query_pos is '错误位置所在的字符计数';
comment on column ${LOG_ANALYSE_TABLE}.location is '在 PostgreSQL 源代码中错误的位置（如果log_error_verbosity被设置为verbose）';
comment on column ${LOG_ANALYSE_TABLE}.application_name is '应用名';
comment on column ${LOG_ANALYSE_TABLE}.backend_type is '后端类型（新加）';
comment on column ${LOG_ANALYSE_TABLE}.leader_pid is 'Leader进程ID（新加）';
comment on column ${LOG_ANALYSE_TABLE}.query_id is '查询ID（新加）';
\q
EOF
read -t 20 -p '按任意键继续......'
echo "##################################################"
echo "导入csv"
while read file
do
echo "----- ${file} -----"
/usr/local/pgsql/bin/psql << EOF
\c ${LOG_ANALYSE_DB}
copy ${LOG_ANALYSE_TABLE} from '${file}' with csv;
\q
EOF
done < ${LOG_FILE_LIST}
echo ""
echo "OK !"
5 最后
爱你
1045
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
