HAWQ外部表

2020/09/08 HAWQ 共 3999 字,约 12 分钟

参考文献

http://www.oushu.io/docs/ch/gpfdist-external-table-import.html?highlight=external http://www.oushu.io/docs/ch/hdfs-external-table-import.html?highlight=external

测试的软件版本

PostgreSQL 8.2.15 
(OushuDB 3.4.0.0) 
(Apache HAWQ 2.4.0.0) 
(Greenplum Database 4.2.0 build 1) 
on x86_64-unknown-linux-gnu, 
compiled by GCC clang version 8.0.1 (tags/RELEASE_801/final) 
compiled on Jan 15 2020 05:46:01

语法

DB=# \h CREATE EXTERNAL TABLE

规范

命名规范

ext_XXXXXX
err_XXXXXX

err表需要定期清理

建议做成存储过程定期清理

建立gpfdist外部表

启动gpfdist服务(文件服务器)

nohup gpfdist -d /home/gpadmin -p 8888 > gpfdist.log 2>&1 &

检查gpfdist服务是否存在(文件服务器)

-bash-4.2$ ps -ef | grep gpfdist
gpadmin  507192 493162  0 01:57 pts/0    00:00:00 gpfdist -d /home/gpadmin -p 8888

建立可写gpfdist外部表

create writable external table ext_w_test001(                             
id bigint, 
name varchar(128)                                                         
)                                                                         
location('gpfdist://10.0.0.6:8888/gpextdata/test001.txt')  
format 'csv'                                                             
encoding 'GB18030';

可写测试

insert into ext_w_test001 select generate_series(1,1000,1),1;

# 插入数据后登陆服务器10.0.0.6(文件服务器)执行下列shell指令 
wc -l /home/gpadmin/gpextdata/test001.txt

建立可读gpfdist外部表

create readable external table ext_r_test001(                             
id bigint, 
name varchar(128)
)                                                                         
location('gpfdist://10.0.0.6:8888/gpextdata/test001.txt')  
format 'csv'                                         
encoding 'GB18030'
log errors into err_r_test001
segment reject limit 10 rows;

可读测试

select count(*) from ext_r_test001;
select * from ext_r_test001 limit 5;

删除gpfdist外部表

找出对应的外部表和err表

方案一

    External table "public.ext_w_test001"
 Column |          Type          | Modifiers 
--------+------------------------+-----------
 id     | bigint                 | 
 name   | character varying(128) | 
Type: writable
Encoding: GB18030
Format type: csv
Format options: delimiter ',' null '' escape '"' quote '"'
External location: gpfdist://10.0.0.6:8888/gpextdata/test001.txt
Distributed randomly

dw=# \d ext_r_test001 
    External table "public.ext_r_test001"
 Column |          Type          | Modifiers 
--------+------------------------+-----------
 id     | bigint                 | 
 name   | character varying(128) | 
Type: readable
Encoding: GB18030
Format type: csv
Format options: delimiter ',' null '' escape '"' quote '"'
External location: gpfdist://10.0.0.6:8888/gpextdata/test001.txt
Segment reject limit: 10 rows
Error table: err_r_test001

方案二

SELECT x.location, 
x.fmttype, 
x.fmtopts, 
x.command,                       
x.rejectlimit, 
x.rejectlimittype, 
x.writable,                             
(SELECT relname 
    FROM pg_class WHERE Oid=x.fmterrtbl) AS errtblname,  
    pg_catalog.pg_encoding_to_char(x.encoding
),
x.fmterrtbl = x.reloid AS errortofile                                     
FROM pg_catalog.pg_exttable x, 
     pg_catalog.pg_class c 
WHERE x.reloid = c.oid 
AND c.oid = 'public.ext_r_test001'::regclass;

删除外部表(删除之前查看建表语法有没有错误表存在)

drop table err_r_test001;
drop external table public.ext_r_test001;

drop external table public.ext_w_test001;

关闭gpfdist服务(文件服务器)

ps -ef | grep gpfdist | grep -v grep | awk '{print $2}' | xargs kill

ps -ef | grep gpfdist

建立web外部表

启动httpd服务(文件服务器)

service httpd start
service httpd status

检查http服务是否可用

echo "1,2" >/var/www/html/text.csv
wget 10.0.0.4:80/text.csv

建立可读web外部表

create readable external web table ext_r_web_test(
id bigint,
name varchar(128)  
)
location ('http://10.0.0.4:80/text.csv')
format 'csv'
encoding 'GB18030'
log errors into err_w_web_test
segment reject limit 10 rows;

可读测试

select * from ext_r_web_test;

关闭httpd服务(文件服务器)

service httpd stop

建立cmd外部表

可写cmd外部表对应脚本准备

echo "cat > /tmp/externala" >/home/gpadmin/unload.sh

建立可写cmd外部表

create writable external web table ext_w_cmd_web(
 id bigint,
 name varchar(128)
 )
execute '/home/gpadmin/unload.sh' on 3
 format 'csv'
 encoding 'GB18030';

可写测试

insert into ext_w_cmd_web select generate_series(1,30,1),1;

## 检查数据文件
gpssh -f all_hosts "cat /tmp/externala"

可读cmd外部表对应脚本准备

echo "cat  /tmp/externala" > /home/gpadmin/load.sh

建立可读cmd外部表

create readable external web table ext_r_cmd_web(
 id bigint,
 name varchar(128)
 )
 execute '/home/gpadmin/load.sh' on 3
 format 'csv'
 encoding 'GB18030'
 log errors into err_r_cmd_tab
 segment reject limit 10 rows;

可读测试

select count(*) from ext_r_cmd_web;
select * from ext_r_cmd_web;

建立hdfs外部表

建立可写hdfs外部表

CREATE writable EXTERNAL TABLE ext_w_hdfs_test
(id int, name text, sponsor text) 
LOCATION ('hdfs://oushu/hawq/default_filespace/test' ) FORMAT 'CSV'
 encoding 'GB18030';

可写测试

insert into ext_w_hdfs_test select 1,2,3;

建立可读hdfs外部表

CREATE readable EXTERNAL TABLE ext_r_hdfs_test
(id int, name text, sponsor text) 
LOCATION ('hdfs://oushu/hawq/default_filespace/test' ) FORMAT 'CSV'
encoding 'GB18030'
log errors into err_r_hdfs_tab
segment reject limit 10 rows;

可读测试

select count(*) from ext_r_hdfs_test;
select * from ext_r_hdfs_test;

文档信息

Search

    Table of Contents