Retry管理

Retry是数据库在SQL或存储过程（包含匿名块）执行失败时，在数据库内部进行重新执行的过程，以提高执行成功率和用户体验。数据库内部通过检查发生错误时的错误码及Retry相关配置，决定是否进行重试。

• 失败时回滚之前执行的语句，并重新执行存储过程进行Retry。

  示例：

  首先要将需要重置的错误码写入到retry配置文件中（gsql同级目录下的retry_errcodes.conf文件），错误码可以通过\set VERBOSITY verbose 后获取:

  openGauss=# \set VERBOSITY verbose
  openGauss=# call retry_basic(1);
  ERROR:  column "b" of relation "t1" does not exist
  SQLSTATE: 42703          ------错误码
  LINE 1: INSERT INTO t1 (b) VALUES (x+1)
                          ^
  QUERY:  INSERT INTO t1 (b) VALUES (x+1)
  CONTEXT:  PL/pgSQL function retry_basic(integer) line 3 at SQL statement
  

  将错误码写入retry_errcodes.conf配置文件可以参考如下命令，并且写完错误码后需要退出和重新进入gsql生效。

  echo "42703" >> ${GAUSSHOME}/bin/retry_errcodes.conf
  

  重试失败的sql示例

  openGauss=# create table t1(a int);
  CREATE TABLE
  openGauss=# CREATE OR REPLACE PROCEDURE retry_basic ( IN  x INT)
  openGauss-# AS
  openGauss$# BEGIN
  openGauss$#   INSERT INTO t1 (a) VALUES (x);
  openGauss$#   INSERT INTO t1 (b) VALUES (x+1);
  openGauss$# END;
  openGauss$# /
  CREATE PROCEDURE
  openGauss=#
  openGauss=# \set RETRY
  Retry is on with default retry times: 5.
  openGauss=#
  openGauss=# call retry_basic(1);
  INFO: query retry 1 time(s).
  INFO: query retry 2 time(s).
  INFO: query retry 3 time(s).
  INFO: query retry 4 time(s).
  INFO: query retry 5 time(s).
  ERROR:  column "b" of relation "t1" does not exist
  LINE 1: INSERT INTO t1 (b) VALUES (x+1)
                          ^
  QUERY:  INSERT INTO t1 (b) VALUES (x+1)
  CONTEXT:  PL/pgSQL function retry_basic(integer) line 3 at SQL statement
  openGauss=#
  openGauss=# select * from t1;
   a
  ---
  (0 rows)
  openGauss=#
  openGauss=# \set RETRY
  Retry is off.
  openGauss=#