文档首页/ 云数据库 GaussDB/ 开发指南（分布式_V2.0-3.x）/ 自治事务/ 规格约束
更新时间：2024-12-19 GMT+08:00
查看PDF
分享

规格约束

  • 自治事务执行时，将会在后台启动自治事务session，我们可以通过max_concurrent_autonomous_transactions设置自治事务执行的最大并行数量，取值范围：0~10000，默认值：10。
  • 当max_concurrent_autonomous_transactions参数设置为0时，自治事务将无法执行。
  • 自治事务新启session后，将使用默认session参数，不共享主session下对象（包括session级别变量、本地临时变量、全局临时表的数据等）。
  • 自治事务理论上限为10000，实际上限为动态值，参考GUC参数max_concurrent_autonomous_transactions描述。
  • 自治事务受通信缓冲区影响，返回给客户端的信息大小受限于通信缓冲区长度，超过通信缓冲区长度时报错。
  • 自治事务执行期间，主事务会给自治事务加锁，异常情况导致此锁残留的场景下，需要依赖gs_clean清理残留锁，清理周期由GUC gs_clean_timeout控制，默认值为60秒。
  • 自治事务设置建立连接超时时间5s，建立连接尝试5次。建立连接期间不立即响应信号，每次建立连接前检查信号。高并发、高CPU、高内存，以及线程池扩容场景下可能存在超时报错现象。
  • 触发器函数不支持自治事务。 
    gaussdb=# CREATE TABLE test_trigger_des_tbl(id1 int, id2 int, id3 int);
NOTICE:  The 'DISTRIBUTE BY' clause is not specified. Using 'id1' as the distribution column by default.
HINT:  Please use 'DISTRIBUTE BY' clause to specify suitable data distribution column.
CREATE TABLE

gaussdb=# CREATE OR REPLACE FUNCTION tri_insert_func() RETURNS TRIGGER AS
$$
DECLARE
 PRAGMA AUTONOMOUS_TRANSACTION;
BEGIN
INSERT INTO test_trigger_des_tbl VALUES(new.id1, new.id2, new.id3);
RETURN new;END$$ LANGUAGE plpgsql;
ERROR:  Triggers do not support autonomous transactions
DETAIL:  N/A

gaussdb=# DROP TABLE test_trigger_des_tbl;
DROP TABLE
  • 自治事务不支持非顶层匿名块调用（仅支持顶层自治事务，包括存储过程、函数、匿名块）。 
    gaussdb=# CREATE TABLE t1(a INT ,b TEXT);
NOTICE:  The 'DISTRIBUTE BY' clause is not specified. Using 'a' as the distribution column by default.
HINT:  Please use 'DISTRIBUTE BY' clause to specify suitable data distribution column.
CREATE TABLE

gaussdb=# DECLARE 
 --PRAGMA AUTONOMOUS_TRANSACTION;
BEGIN
 DECLARE 
  PRAGMA AUTONOMOUS_TRANSACTION;
 BEGIN
  DBE_OUTPUT.PRINT_LINE('JUST USE CALL.');
  INSERT INTO t1 VALUES(1,'CAN YOU ROLLBACK!');
 END;
 INSERT INTO t1 VALUES(2,'I WILL ROLLBACK!');
 ROLLBACK;
END;
/
JUST USE CALL.
ANONYMOUS BLOCK EXECUTE

gaussdb=# SELECT * FROM t1;
 a | b 
---+---
(0 rows)

gaussdb=# DROP TABLE t1;
DROP TABLE
  • 自治事务不支持ref_cursor参数传递。 
    gaussdb=# create table sections(section_ID int);
gaussdb=# insert into sections values(1);
gaussdb=# insert into sections values(1);
gaussdb=# insert into sections values(1);
gaussdb=# insert into sections values(1);

gaussdb=# CREATE OR REPLACE function proc_sys_ref()
return SYS_REFCURSOR
IS
declare
  PRAGMA AUTONOMOUS_TRANSACTION;
        C1 SYS_REFCURSOR;
BEGIN
 OPEN C1 FOR SELECT section_ID FROM sections ORDER BY section_ID;
 return C1;
END;
/
ERROR:  Autonomous function do not support ref cursor as return types or out, inout arguments.
DETAIL:  N/A
gaussdb=# CREATE OR REPLACE function proc_sys_ref(OUT C2 SYS_REFCURSOR, OUT a int)
return SYS_REFCURSOR
IS
declare
  PRAGMA AUTONOMOUS_TRANSACTION;
        C1 SYS_REFCURSOR;
BEGIN
 OPEN C1 FOR SELECT section_ID FROM sections ORDER BY section_ID;
 return C1;
END;
/
ERROR:  Autonomous function do not support ref cursor as return types or out, inout arguments.
DETAIL:  N/A
  • 分布式自治事务不支持下推（IMMUTABLE、STABLE类型）。 
    gaussdb=# CREATE OR REPLACE PROCEDURE autonomous_test_in_p_116(num1 INT ) 
IMMUTABLE 
AS
DECLARE
PRAGMA AUTONOMOUS_TRANSACTION;
BEGIN
PERFORM pg_sleep(1);
END;
/
ERROR:  Autonomous transactions do not support STABLE/IMMUTABLE.
DETAIL:  Please remove stable/immutable.

gaussdb=# CREATE OR REPLACE PROCEDURE autonomous_test_in_p_117(num1 INT ) STABLE AS
DECLARE
PRAGMA AUTONOMOUS_TRANSACTION;
BEGIN
PERFORM pg_sleep(1);
END;
/
ERROR:  Autonomous transactions do not support STABLE/IMMUTABLE.
DETAIL:  Please remove stable/immutable.
  • 分布式不支持检测（死锁时，有锁等待超时报错）。 
    gaussdb=# CREATE TABLE test_lock (id INT,a DATE);
NOTICE:  The 'DISTRIBUTE BY' clause is not specified. Using 'id' as the distribution column by default.
HINT:  Please use 'DISTRIBUTE BY' clause to specify suitable data distribution column.
CREATE TABLE
gaussdb=# INSERT INTO test_lock VALUES (10,SYSDATE),(11,SYSDATE),(12,SYSDATE);
INSERT 0 3
gaussdb=# CREATE OR REPLACE FUNCTION autonomous_test_lock(num1 INT,num2 INT) RETURNS
INTEGER LANGUAGE plpgsql AS $$
DECLARE num3 INT := 4;
PRAGMA AUTONOMOUS_TRANSACTION;
BEGIN
UPDATE test_lock SET a=SYSDATE WHERE id =11;
RETURN num1+num2+num3;
END;
$$;
CREATE FUNCTION

gaussdb=# START TRANSACTION;
START TRANSACTION
gaussdb=# UPDATE test_lock SET a=SYSDATE WHERE id =11;
UPDATE 1
gaussdb=# CALL autonomous_test_lock(1,1);
ERROR:  ERROR:  Lock wait timeout: thread 139874535470848 on node datanode1 waiting for ShareLock on transaction 16214 after 120000.124 ms
DETAIL:  blocked by hold lock thread 139874577413888, statement <UPDATE test_lock SET a = "sysdate"() WHERE id =11;>, hold lockmode ExclusiveLock.
CONTEXT:  SQL statement "UPDATE test_lock SET a=SYSDATE WHERE id =11"
PL/SQL function autonomous_test_lock(integer,integer) line 5 at SQL statement
referenced column: autonomous_test_lock

gaussdb=# END;
ROLLBACK
gaussdb=# DROP TABLE test_lock;
DROP TABLE
  • 自治事务函数不支持直接返回record类型和out出参，与record类型同时返回。 
    gaussdb=# CREATE OR REPLACE FUNCTION auto_func() RETURN RECORD
AS
DECLARE
  TYPE rec_type IS RECORD(c1 INT, c2 INT);
  r rec_type;
  PRAGMA AUTONOMOUS_TRANSACTION;
BEGIN
  r.c1:=101;
  r.c2:=201;
  RETURN r;
END;
/
CREATE FUNCTION

gaussdb=# SELECT auto_func();
ERROR:  unrecognized return type for PLSQL function.
CONTEXT:  referenced column: auto_func
  • 不支持修改自治事务的隔离级别。 
    gaussdb=# CREATE OR REPLACE PROCEDURE auto_func(r INT)
AS
DECLARE
  a INT;
  PRAGMA AUTONOMOUS_TRANSACTION;
BEGIN
SET TRANSACTION ISOLATION LEVEL SERIALIZABLE;
 a:=r;
END;
/
CREATE FUNCTION
gaussdb=# call auto_func(1);
ERROR:  ERROR:  SET TRANSACTION ISOLATION LEVEL must be called before any query
CONTEXT:  SQL statement "SET TRANSACTION ISOLATION LEVEL SERIALIZABLE"
PL/SQL function auto_func(integer) line 6 at SQL statement
referenced column: auto_func
  • 不支持自治事务返回集合类型（setof）。 
    gaussdb=# CREATE OR REPLACE FUNCTION test_set() RETURN SETOF INT
AS
DECLARE
PRAGMA AUTONOMOUS_TRANSACTION;
BEGIN
    RETURN QUERY (SELECT unnest(ARRAY[ARRAY[1, 2], ARRAY[3, 4]]));
END;
/
ERROR:  Autonomous transactions do not support RETURN SETOF.
DETAIL:  N/A
父主题： 自治事务

相关文档