本文为您介绍如何使用gist 索引。 注意 此功能需加载插件pgtrgm。 plaintext teledb create extension pgtrgm; CREATE EXTENSION teledb create table ttrgm (id int,trgm text,notrgm text) ; ERROR: relation "ttrgm" already exists teledb create index ttrgmtrgmidx on ttrgm using gist(trgm gisttrgmops); CREATE INDEX

本文介绍如何在使用SELECT语法时进行聚集查询，例如统计记录数、不重复值的记录表，以及求和等。 统计记录数。 plaintext teledb select count(1) from teledbpg; count 5 (1 row) 统计不重复值的记录表。 plaintext teledb select count(distinct id) from teledbpg; count 4 (1 row) 求和。 plaintext teledb select sum(id) from teledbpg; sum 11 (1 row) 求最大值。 plaintext teledb select max(id) from teledbpg; max 4 (1 row) 求最小值。 plaintext teledb select min(id) from teledbpg; min 1 (1 row) 求平均值。 plaintext teledb select avg(id) from teledbpg; avg 2.2000000000000000 (1 row)

添加外键 plaintext teledb create table tp(f1 int not null,f2 int ,primary key(f1)); CREATE TABLE teledb create table tf(f1 int not null,f2 int ); CREATE TABLE teledb ALTER TABLE tf ADD CONSTRAINT tff1fkey FOREIGN KEY (f1) REFERENCES tp (f1); ALTER TABLE teledb d+ tf Table "public.tf" Column Type Collation Nullable Default Storage Stats target Description +++++++ f1 integer not null plain f2 integer plain Foreignkey constraints: "tff1fkey" FOREIGN KEY (f1) REFERENCES tp(f1) Distribute By: SHARD(f1) Location Nodes: ALL DATANODES 外键使用限制： 外键只是同一个节点内约束有效果，所以外键字段和对应主键字段必需都是表的分布键，否则由于数据分布于不同的节点内会导致更新失败。 分区表和冷热分区表也不支持外键，数据分区后位于不同的物理文件中，无法约束。 删除外键 plaintext teledb ALTER TABLE tf DROP CONSTRAINT tff1fkey; ALTER TABLE 修改表所属模式 plaintext teledb create schema teledb; CREATE SCHEMA teledb dt t List of relations Schema Name Type Owner +++ public t table teledb (1 row) teledb alter table t set schema teledb; ALTER TABLE teledb dt t List of relations Schema Name Type Owner +++ teledb t table teledb (1 row) 修改表所属用户 plaintext teledb dt t List of relations Schema Name Type Owner +++ teledb t table teledb (1 row) teledb

select from teledbpg where exists(select from person where person.id teledbpg.id); id nickname + 1 teledb 1 hello,pgxc 2 TELEDB 3 pg (4 rows) exists 等价写法 plaintext teledb select teledbpg. from teledbpg, (select distinct id from person) as t where t.id teledbpg.id; id nickname + 1 teledb 1 hello,pgxc 2 TELEDB 3 pg (4 rows)

本文为您介绍如何管理range分区表、list分区表和多级分区表。 range分区表 创建主分区 plaintext teledb create table tnativerange (f1 bigint,f2 timestamp default now(), f3 integer) partition by range ( f2 ) distribute by shard(f1); CREATE TABLE 建立两个子表 plaintext teledb create table tnativerange201709 partition of tnativerange (f1 ,f2 , f3 ) for values from ('20170901') to ('20171001'); CREATE TABLE teledb create table tnativerange201710 partition of tnativerange (f1 ,f2 , f3 ) for values from ('20171001') to ('20171101'); CREATE TABLE 查看表结构 plaintext teledb d+ tnativerange Table "public.tnativerange" Column Type Collation Nullable Default Storage Stats target Description +++++++ f1 bigint plain f2 timestamp without time zone now() plain f3 integer plain Partition key: RANGE (f2) Partitions: tnativerange201709 FOR VALUES FROM ('20170901 00:00:00') TO ('20171001 00:00:00'), tnativerange201710 FOR VALUES FROM ('20171001 00:00:00') TO ('20171101 00:00:00') Distribute By: SHARD(f1) Location Nodes: ALL DATANODES 创建default分区 不创建default 分区，插入范围越界出错。 plaintext teledb insert into tnativerange values(1,'20160901',1); ERROR: node:dn01, backendpid:39912, nodename:dn01,backendpid:39912,message:no partition of relation "tnativerange" found for row DETAIL: Partition key of the failing row contains (f2) (20160901 00:00:00). 创建default 分区后能正常插入数据。 plaintext teledb

select from teledbpg order by md5(nickname); id nickname + 1 hello,pgxc 3 pg 1 teledb (3 rows) 排序也能用子查询。 plaintext teledb select from teledbpg order by (select id from teledbpg order by random() limit 1); id nickname + 1 teledb 1 hello,pgxc 3 pg (3 rows) null 值排序结果处理 plaintext teledb insert into teledbpg values(4,null); INSERT 0 1 null 值记录排在最前面。 plaintext teledb select from teledbpg order by nickname nulls first; id nickname + 4 1 hello,pgxc 3 pg 1 teledb (4 rows) null 值记录排在最后。 plaintext teledb select from teledbpg order by nickname nulls last; id nickname + 1 hello,pgxc 3 pg 1 teledb 4 (4 rows) 按拼音排序 使用convert 函数实现汉字按拼音进行排序。 plaintext teledb select from (values ('张三'), ('李四'),('陈五')) t(myname) order by convert(myname::bytea,'UTF8','GBK'); myname 陈五 李四 张三 (3 rows) 使用convertto 函数实现汉字按拼音进行排序。 plaintext teledb select from (values ('张三'), ('李四'),('陈五')) t(myname) order by convertto(myname,'GBK'); myname 陈五 李四 张三 (3 rows) 通过指定排序规则collact 来实现汉字按拼音进行排序。 plaintext teledb

本文介绍如何在使用SELECT语法时进行聚合函数并发计算。 plaintext teledb set maxparallelworkerspergather to 0; SET Time: 0.214 ms teledb select count(1) from tcount; count 20000000 (1 row) Time: 2700.569 ms (00:02.701) teledb set maxparallelworkerspergather to 2; SET Time: 0.269 ms teledb select count(1) from tcount; count 20000000 (1 row) Time: 1867.719 ms (00:01.868)

本文为您介绍如何管理视图，包括创建视图、删除视图和物化视图使用。 创建视图 plaintext teledb create view trangeview as select from trange; CREATE VIEW teledb select from trangeview; f1 f2 f3 ++ 1 20230822 11:06:16.9783 1 2 20230822 11:06:16.9783 50 2 20230822 11:06:16.9783 110 3 20230822 11:06:16.9783 100 (4 rows) 数据类型重定义。 plaintext teledb drop view trangeview; DROP VIEW teledb create view trangeview as select f1,f2::date from trange; CREATE VIEW teledb select from trangeview; f1 f2 + 1 20230822 3 20230822 2 20230822 2 20230822 (4 rows) 数据类型重定义，以及取别名。 plaintext teledb drop view trangeview; DROP VIEW teledb create view trangeview as select f1,f2::date as mydate from trange; CREATE VIEW teledb select from trangeview; f1 mydate + 1 20230822 2 20230822 2 20230822 3 20230822 (4 rows) TeleDB支持视图引用表或字段改名联动，不受影响。 plaintext teledb d+ trangeview View "teledb.trangeview" Column Type Collation Nullable Default Storage Description ++++++ f1 bigint plain mydate date plain View definition: SELECT trange.f1, trange.f2::date AS mydate FROM trange; teledb

id nickname + 1 teledb

id nickname + 2 TELEDB 4 1 teledb 1 hello,pgxc

drop index tsexsexidx; DROP INDEX teledb create index tsexsexidx on tsex (sex) where sex '女'; CREATE INDEX teledb analyze tsex; ANALYZE teledb explain select from tsex where sex '男'; QUERY PLAN Remote Fast Query Execution (cost0.00..0.00 rows0 width0) Node/s: dn01, dn02 > Seq Scan on tsex (cost0.00..9966.58 rows500539 width8) Filter: (sex '男'::text) (4 rows) teledb explain select from tsex where sex '女'; QUERY PLAN Remote Fast Query Execution (cost0.00..0.00 rows0 width0) Node/s: dn01, dn02 > Index Scan using tsexsexidx on tsex (cost0.14..13.15 rows67 width8) (3 rows) teledb q [teledb@localhost bin]$ ./telesql p 22222 U teledb d teledb Password for user teledb: telesql (TeleDB V6) Type "help" for help. teledb di+ tsexsexidx List of relations Schema Name Type Owner Table Size Allocated Size Description +++++++ public tsexsexidx index teledb tsex 16 kB 16 kB (1 row)

本文为您介绍如何创建唯一索引。 创建唯一索引 plaintext teledb create unique index tasiduidx on tas using btree(id); CREATE INDEX 非shard key 字段不能建立唯一索引 plaintext teledb create unique index tasiduidx on tas using btree(mc); ERROR: Unique index of partitioned table must contain the hash/modulo distribution column.

d+ tnativemullistgd Table "public.tnativemullistgd" Column Type Collation Nullable Default Storage Stats target Description ++++++ + f1 bigint not null nextval('tnativemullistf1seq'::regclass) plain f2 integer plain f3 text extended f4 text extended f5 timestamp(0) without time zone plain Partition of: tnativemullist FOR VALUES IN ('广东') Partition constraint: ((f3 IS NOT NULL) AND (f3 ANY (ARRAY['广东'::text]))) Partition key: RANGE (f5) Partitions: tnativemullistgd201701 FOR VALUES FROM ('20170101 00:00:00') TO ('20170201 00:00:00'), tnativemullistgd201702 FOR VALUES FROM ('20170201 00:00:00') TO ('20170301 00:00:00') Distribute By: SHARD(f1) Location Nodes: ALL DATANODES TeleDB支持存在1级和2级分区混用，大家不需要都平级。

id nickname + 3 pg

本文为您介绍如何创建分区表。 创建范围分区表 plaintext teledb create table trange (f1 bigint,f2 timestamp default now(), f3 integer) partition by range (f3) begin (1) step (50) partitions (3) distribute by shard(f1); CREATE TABLE teledb insert into trange(f1,f3) values(1,1),(2,50),(3,100),(2,110); COPY 4 teledb insert into trange(f1,f3) values(1,1),(2,50),(3,100),(2,110); COPY 4 teledb d+ trange; Table "public.trange" Column Type Collation Nullable Default Storage Stats target Description +++++++ f1 bigint plain f2 timestamp without time zone now() plain f3 integer plain Distribute By: SHARD(f1) Location Nodes: ALL DATANODES Partition By: RANGE(f3) Of Partitions: 3 Start With: 1 Interval Of Partition: 50 创建时间范围分区表 plaintext teledb create table ttimerange (f1 bigint, f2 timestamp ,f3 bigint) partition by range (f2) begin (timestamp without time zone '20170901 0:0:0') step (interval '1 month') partitions (12) distribute by shard(f1); CREATE TABLE teledb d+ ttimerange Table "public.ttimerange" Column Type Collation Nullable Default Storage Stats target Description +++++++ f1 bigint plain f2 timestamp without time zone plain f3 bigint plain Distribute By: SHARD(f1) Location Nodes: ALL DATANODES Partition By: RANGE(f2)

d dbgbk Did not find any relation named "dbgbk". teledb l+ dbgbk List of databases Name Owner Encoding Collate Ctype Access privileges Size Tablespace Description ++++++++ dbgbk teledb GBK zhCN.gbk zhCN.gbk 19 MB pgdefault (1 row) 指定排序规则 plaintext teledb create database teledbdblccollate lccollate 'C' template template0; CREATE DATABASE 指定分组规则 plaintext teledb create database teledblcctype LCCTYPE 'C' template template0; CREATE DATABASE 配置数据可连接 plaintext teledb create database teledballowconnections ALLOWCONNECTIONS true; CREATE DATABASE teledb select datallowconn from pgdatabase where datname 'teledballowconnections'; datallowconn t (1 row) teledb c teledballowconnections You are now connected to database "teledballowconnections" as user "teledb". 配置连接数 plaintext teledb create database teledbconnlimit CONNECTION LIMIT 100; CREATE DATABASE teledb select datconnlimit from pgdatabase where datname'teledbconnlimit'; datconnlimit 100 (1 row) 配置数据库可以被复制（是否模板数据库） plaintext teledb create database teledbistemplate istemplate true; CREATE DATABASE teledb select datconnlimit from pgdatabase where datname'teledbconnlimit'; datconnlimit 100 (1 row) 多个参数一起配置 plaintext teledb create database teledbmul owner teledbuser CONNECTION LIMIT 50 template template0 encoding 'utf8' lccollate 'C'; CREATE DATABASE

id name birth city +++ 1 张三 20001201 00:00:00 北京 2 李四 19970324 00:00:00 上海 3 王五 20040901 00:00:00 广州

Of Partitions: 12 Start With: 20170901 Interval Of Partition: 1 MONTH 说明 partion by range(xx) 代表范围分区， 支持timesamp,int类型，数据分布于那个子表就是根据这个字段值来计算分区。 begin(xx) 表示开始分区的初始值。 partition(xx) 代表初始分区数量。 step (xx) 代表分区步长。

集群监控页面展示整个集群监控指标，可以从实例的整体全局上了解实例的运行状态。 操作场景 集群监控页面展示整个集群监控指标，可以从实例的整体全局上了解实例的运行状态。 操作步骤 1. 以用户名和密码登录分布式数据库 控制台，在左侧导航树单击实例监控 ，选择集群监控 页签。 2. 选择需要查看的时间范围。 集群监控页面包含了以下指标信息： 容量占用率：所有节点数据目录大小之和 / 所有节点磁盘总量之和。 总请求数：所有DN节点总请求数之和。 读请求数：所有DN节点读请求数之和。 写请求数：所有DN节点写请求数之和。 其他请求数：所有DN节点其他请求数之和。 总连接数：所有DN节点连接数之和。

truncate ttimerange partition for ('20170901' ::timestamp without time zone); TRUNCATE TABLE teledb select from ttimerange; f1 f2 f3 ++ 1 20171001 00:00:00 100 1 20171101 00:00:00 100 (2 rows) truncate 一个数字分区表 plaintext teledb select from trange; f1 f2 f3 ++ 1 20230823 10:29:46.263768 1 2 20230823 10:29:46.263768 50 2 20230823 10:29:46.263768 110 3 20230823 10:29:46.263768 100 (4 rows) teledb d+ trange Table "public.trange" Column Type Collation Nullable Default Storage Stats target Description +++++++ f1 bigint plain f2 timestamp without time zone now() plain f3 integer plain Distribute By: SHARD(f1) Location Nodes: ALL DATANODES Partition By: RANGE(f3) Of Partitions: 3 Start With: 1 Interval Of Partition: 50 teledb truncate trange partition for (1); TRUNCATE TABLE teledb truncate trange partition for (2); TRUNCATE TABLE teledb select from trange; f1 f2 f3 ++ 2 20230823 10:29:46.263768 110 3 20230823 10:29:46.263768 100 (2 rows)

本文为您介绍如何在使用SELECT语法时添加where条件。 单条件查询 plaintext teledb select from teledbpg where id 1; id nickname + 1 teledb 1 hello,pgxc (2 rows) 多条件and plaintext teledb select from teledbpg where id 1 and nickname like '%h%' ; id nickname + 1 hello,pgxc (1 row) 多条件or plaintext teledb select from teledbpg where id 1 or nickname like '%p%' ; id nickname + 1 teledb 1 hello,pgxc 3 pg (3 rows) ilike 不区分大小写匹配 plaintext teledb insert into teledbpg values(2,'TELEDB'); INSERT 0 1 teledb select from teledbpg where nickname ilike '%te%'; id nickname + 1 teledb 2 TELEDB (2 rows) where 条件也能支持子查询 plaintext teledb select from teledbpg where id(select (random())::integer from teledbpg order by random() limit 1); id nickname + (0 rows) teledb select from teledbpg where id(select (random())::integer from teledbpg order by random() limit 1); id nickname + 1 teledb 1 hello,pgxc (2 rows) null 值查询方法 plaintext teledb select from teledbpg where nickname is null; id nickname + 4 (1 row) teledb select from teledbpg where nickname is not null; id nickname + 1 teledb 3 pg 1 hello,pgxc 2 TELEDB (4 rows) exists 只要有记录返回就为真 plaintext teledb

explain select from tmulidx where f11 or f22 or f33 ; QUERY PLAN Remote Fast Query Execution (cost0.00..0.00 rows0 width0) Node/s: dn001, dn002 > Seq Scan on tmulidx (cost0.00..12966.87 rows3 width16) Filter: ((f1 1) OR (f2 2) OR (f3 3)) (4 rows) Time: 3.153 ms 如果返回字段全部在索引文件中，则只需要扫描索引，IO 开销会更少。 plaintext teledb explain select f1,f2,f3 from tmulidx where f11 ; QUERY PLAN Remote Fast Query Execution (cost0.00..0.00 rows0 width0) Node/s: dn01 > Index Only Scan using tmulidxidx on tmulidx (cost0.42..8.44 rows1 width12) Index Cond: (f1 1) (4 rows) Time: 2.630 ms 更新性能比单字段多索引文件要好 多字段 plaintext teledb truncate table tmulidx; TRUNCATE TABLE Time: 44.930 ms teledb insert into tmulidx select t,t,t,t from generateseries(1,1000000) as t; INSERT 0 1000000 Time: 1985.901 ms (00:01.986) 单字段 plaintext teledb create table tsimpleidx (f1 int,f2 int,f3 int,f4 int); CREATE TABLE teledb create index tsimpleidxidx on tsimpleidx(f1); CREATE INDEX teledb create index tsimpleidxidx1 on tsimpleidx(f2); CREATE INDEX teledb create index tsimpleidxidx2 on tsimpleidx(f3); CREATE INDEX teledb

该任务用于指导软件工程师安装部署teledb管控服务。 该任务用于指导软件工程师安装部署teledb管控服务。 前提条件 已初始化环境，包括创建teledb用户，挂载磁盘，并给于sudo权限。 操作步骤 以下操作以x86安装包为例： 1. 创建根目录提供给挂载盘/data使用，并给于sudo权限。 plaintext mkdir p /data mount /dev/vdb /data echo "UUID"blkid /dev/vdbcut d '"' f2" /data xfs defaults 0 0" sudo tee a /etc/fstab 2. 上传并解压安装包 plaintext cd /app 上传teledbdeployx86centos.zip和teledbxossxxxi.x8664.rpm至app目录下。 unzip teledbdeployx86centos.zip cd teledbdeployx86centos/teledbxmirror/pgxzm cp teledbxossxxxi.x8664.rpm 3. 安装dos2unix 执行如下命令 plaintext cd /app/teledbdeployx86centos/teledbxmirror/deploy/utils/ rpm ivh dos2unix6.0.34.el7.x8664.rpm 出现如下回显信息。 plaintext cd /app chmod R 755 teledbdeployx86centos chown R teledb:teledb /app 4. 修改配置文件role.info和ossinit.conf。 执行如下命令修改role.info配置文件。 plaintext su teledb cd /app/teledbdeployx86centos/teledbxmgr/conf vim role.info 您可按照如下模版修改配置文件内容： 需要修改的就是主机ip、ssh端口、ssh连接用户（teledb）、用户密码

xcnodeid nodename id nickname +++ 1085152094 dn01 5 test 1085152094 dn01 6 teledb, teledb1 506537247 dn02 3 pg

id nickname + 2 TELEDB 4

id nickname + 4

id nickname + 3 pg 4 2 TELEDB

本文为你介绍合并多个查询结果的方法。 不过滤重复的记录。 plaintext teledb create table teledbpg1(id int, nickname varchar); CREATE TABLE teledb insert into teledbpg1 values(3, 'pg'),(5,'test'); COPY 2 teledb select from teledbpg union all select from teledbpg1; id nickname + 1 teledb 1 hello,pgxc 2 TELEDB 3 pg 4 5 test 3 pg (7 rows) 过滤重复的记录。 plaintext teledb select from teledbpg union select from teledbpg1; id nickname + 1 teledb 1 hello,pgxc 4 5 test 2 TELEDB 3 pg (6 rows) 每个子查询分布在合并结果中的使用。 plaintext teledb select from (select from teledbpg limit 2) as t union all select from (select from teledbpg1 limit 2) teledb ; id nickname + 1 teledb 1 hello,pgxc 5 test 3 pg (4 rows)

本文为您介绍如何管理普通索引。 plaintext teledb create index tididx on t1 using btree(id); CREATE INDEX

本文为您介绍如何删除索引。 plaintext teledb drop index tmulidxidx; DROP INDEX

Column Type Collation Nullable Default Storage Stats target Description +++++++ id integer plain mc text extended 这是一条字段注释 age integer plain

本文为您介绍如何创建唯一索引。 创建唯一索引 plaintext teledb create unique index tasiduidx on tas using btree(id); CREATE INDEX 非shard key 字段不能建立唯一索引 plaintext teledb create unique index tasiduidx on tas using btree(mc); ERROR: Unique index of partitioned table must contain the hash/modulo distribution column.