JBPM数据清理

接手一个JBPM3项目, MYSQL INNODB ibdata1都50GB了, 关联查询processdefinition, processintance很慢, 需要清楚一些过期的processinstance.

使用GraphSession.deleteProcessInstance删? 一两秒才能删一个, 几十万条, outage不允许.

网上一哥们使用delete from ... where processinstance in (...) 批量的删, 慢到你不信, SET FOREIGN_KEY_CHECKS=0;也不好使.

jbpm_token 70万条, jbpm_variableinstance 700万条, jbpm_bytearray, jbpm_log, jbpm_byteblock好像都是千万条的级别.

用户要求保留近期的数据不能直接drop掉.

硬着头皮用power designer导入这些jbpm_表慢慢分析这些关联.

DROP TABLE IF EXISTS TMP_PROCESSINSTANCE; DROP TABLE IF EXISTS TMP_TOKEN; DROP TABLE IF EXISTS TMP_MODULEINSTANCE; DROP TABLE IF EXISTS TMP_SWIMLANEINSTANCE; DROP TABLE IF EXISTS TMP_TOKENVARIABLEMAP; DROP TABLE IF EXISTS TMP_VARIABLEINSTANCE; DROP TABLE IF EXISTS TMP_TASKINSTANCE; DROP TABLE IF EXISTS TMP_TASKACTORPOOL; DROP TABLE IF EXISTS TMP_LOG; DROP TABLE IF EXISTS TMP_BYTEARRAY; select * from jbpm_processdefinition into outfile 'c:/jbpm_processdefinition.sql'; select * from `jbpm_processinstance` pi where pi.`end_` > '2009-08-01 00:00:00' into outfile 'c:/jbpm_processinstance.sql'; #Assume we create a tmp table named tmp_processinstance create table TMP_PROCESSINSTANCE (ID_ bigint not null auto_increment, VERSION_ integer not null, START_ datetime, END_ datetime, ISSUSPENDED_ bit, PROCESSDEFINITION_ bigint, ROOTTOKEN_ bigint, SUPERPROCESSTOKEN_ bigint, primary key (ID_)) type=InnoDB; load data infile 'c:/jbpm_processinstance.sql' into table TMP_PROCESSINSTANCE; #Each processinstance owns one roottoken and so roottoken is not a condition select t.* from `jbpm_token` t,TMP_PROCESSINSTANCE pi where t.`PROCESSINSTANCE_`=pi.`ID_` or t.`ID_`=pi.`ROOTTOKEN_` or t.`SUBPROCESSINSTANCE_`=pi.`ID_` or t.`ID_`=pi.`SUPERPROCESSTOKEN_` into outfile 'c:/jbpm_token.sql'; #Assume the table name is tmp_token create table TMP_TOKEN (ID_ bigint not null auto_increment, VERSION_ integer not null, NAME_ varchar(255), START_ datetime, END_ datetime, NODEENTER_ datetime, NEXTLOGINDEX_ integer, ISABLETOREACTIVATEPARENT_ bit, ISTERMINATIONIMPLICIT_ bit, ISSUSPENDED_ bit, NODE_ bigint, PROCESSINSTANCE_ bigint, PARENT_ bigint, SUBPROCESSINSTANCE_ bigint, primary key (ID_)) type=InnoDB; load data infile 'c:/jbpm_token.sql' into table TMP_TOKEN; select * from `jbpm_timer` t where t.`PROCESSINSTANCE_` in ( select pi.`ID_` from TMP_PROCESSINSTANCE pi ) into outfile 'c:/jbpm_timer.sql'; select * from `jbpm_moduleinstance` mi where mi.`PROCESSINSTANCE_` IN( select pi.`ID_` from TMP_PROCESSINSTANCE pi ) into outfile 'c:/jbpm_moduleinstance.sql'; #Assume we create tmp_moduleinstance table. create table TMP_MODULEINSTANCE (ID_ bigint not null auto_increment, CLASS_ char(1) not null, PROCESSINSTANCE_ bigint, TASKMGMTDEFINITION_ bigint, NAME_ varchar(255), primary key (ID_)) type=InnoDB; load data infile 'c:/jbpm_moduleinstance.sql' into table TMP_MODULEINSTANCE; select * from `jbpm_runtimeaction` ra where ra.`PROCESSINSTANCE_` in ( select pi.`ID_` from TMP_PROCESSINSTANCE pi ) into outfile 'c:/jbpm_runtimeaction.sql'; select * from `jbpm_action` into outfile 'c:/jbpm_action.sql'; select * from `jbpm_variableaccess` into outfile 'c:/jbpm_variableaccess.sql'; select * from `jbpm_delegation` into outfile 'c:/jbpm_delegation.sql'; select * from `jbpm_event` into outfile 'c:/jbpm_event.sql'; select * from `jbpm_node` into outfile 'c:/jbpm_node.sql'; select * from `jbpm_message` into outfile 'c:/jbpm_message.sql'; select * from `jbpm_exceptionhandler` into outfile 'c:/jbpm_exceptionhandler.sql'; select * from `jbpm_taskcontroller` into outfile 'c:/jbpm_taskcontroller.sql'; select * from `jbpm_task` into outfile 'c:/jbpm_task.sql'; select * from `jbpm_moduledefinition` into outfile 'c:/jbpm_moduledefinition.sql'; select * from `jbpm_swimlane` into outfile 'c:/jbpm_swimlane.sql'; select * from `jbpm_decisionconditions` into outfile 'c:/jbpm_decisionconditions.sql'; select * from `jbpm_transition` into outfile 'c:/jbpm_transition.sql'; #Instances records. select * from `jbpm_swimlaneinstance` si where si.`TASKMGMTINSTANCE_` IN( select mi.`ID_` from `tmp_moduleinstance` mi ) into outfile 'c:/jbpm_swimlaneinstance.sql'; #Assume we create tmp table tmp_swimlaneinstance create table TMP_SWIMLANEINSTANCE (ID_ bigint not null auto_increment, NAME_ varchar(255), ACTORID_ varchar(255), SWIMLANE_ bigint, TASKMGMTINSTANCE_ bigint, primary key (ID_)) type=InnoDB; load data infile 'c:/jbpm_swimlaneinstance.sql' into table TMP_SWIMLANEINSTANCE; select * from `jbpm_tokenvariablemap` tm where tm.`TOKEN_` IN ( select t.`ID_` from `tmp_token` t ) or tm.`CONTEXTINSTANCE_` IN ( select mi.`ID_` from `tmp_moduleinstance` mi ) into outfile 'c:/jbpm_tokenvariablemap.sql'; #Assume we create table tmp_tokenvariablemap create table TMP_TOKENVARIABLEMAP (ID_ bigint not null auto_increment, TOKEN_ bigint, CONTEXTINSTANCE_ bigint, primary key (ID_)) type=InnoDB; load data infile 'c:/jbpm_tokenvariablemap.sql' into table TMP_TOKENVARIABLEMAP; select * from `jbpm_variableinstance` vi where vi.`PROCESSINSTANCE_` IN( select pi.`ID_` from tmp_processinstance pi ) or vi.`TOKEN_` IN ( select t.`ID_` from tmp_token t ) or vi.`TOKENVARIABLEMAP_` IN ( select `ID_` from `tmp_tokenvariablemap` ) into outfile 'c:/jbpm_variableinstance.sql'; #Assume we create temp table tmp_variableinstance create table TMP_VARIABLEINSTANCE (ID_ bigint not null auto_increment, CLASS_ char(1) not null, NAME_ varchar(255), CONVERTER_ char(1), TOKEN_ bigint, TOKENVARIABLEMAP_ bigint, PROCESSINSTANCE_ bigint, BYTEARRAYVALUE_ bigint, DATEVALUE_ datetime, DOUBLEVALUE_ double precision, LONGIDCLASS_ varchar(255), LONGVALUE_ bigint, STRINGIDCLASS_ varchar(255), STRINGVALUE_ text, TASKINSTANCE_ bigint, primary key (ID_)) type=InnoDB; load data infile 'c:/jbpm_variableinstance.sql' into table TMP_VARIABLEINSTANCE; select * from `jbpm_taskinstance` ti where ti.`TOKEN_` IN ( select `ID_` from `tmp_token` ) or ti.`SWIMLANINSTANCE_` IN ( select `ID_` from `tmp_swimlaneinstance` ) or ti.`TASKMGMTINSTANCE_` IN ( select `ID_` from `tmp_moduleinstance` ) into outfile 'c:/jbpm_taskinstance.sql'; #Assume we create temp table tmp_taskinstance create table TMP_TASKINSTANCE (ID_ bigint not null auto_increment, CLASS_ char(1) not null, NAME_ varchar(255), DESCRIPTION_ text, ACTORID_ varchar(255), CREATE_ datetime, START_ datetime, END_ datetime, DUEDATE_ datetime, PRIORITY_ integer, ISCANCELLED_ bit, ISSUSPENDED_ bit, ISOPEN_ bit, ISSIGNALLING_ bit, ISBLOCKING_ bit, TASK_ bigint, TOKEN_ bigint, SWIMLANINSTANCE_ bigint, TASKMGMTINSTANCE_ bigint, primary key (ID_)) type=InnoDB; load data infile 'c:/jbpm_taskinstance.sql' into table TMP_TASKINSTANCE; select * from `jbpm_comment` c where c.`TOKEN_` IN ( select `ID_` from `tmp_token` ) or c.`TASKINSTANCE_` IN ( select `ID_` from `tmp_taskinstance` ) into outfile 'c:/jbpm_comment.sql'; select * from `jbpm_taskactorpool` tap where tap.`TASKINSTANCE_` IN ( select `ID_` from `tmp_taskinstance` ) into outfile 'c:/jbpm_taskactorpool.sql'; create table TMP_TASKACTORPOOL (TASKINSTANCE_ bigint not null, POOLEDACTOR_ bigint not null, primary key (TASKINSTANCE_, POOLEDACTOR_)) type=InnoDB; load data infile 'c:/jbpm_taskactorpool.sql' into table TMP_TASKACTORPOOL; select * from `jbpm_pooledactor` pa where pa.`ID_` IN ( select `POOLEDACTOR_` from TMP_TASKACTORPOOL ) into outfile 'c:/jbpm_pooledactor.sql'; #FIXME# select * from `jbpm_log` lo where lo.`DATE_` > '2009-08-01 00:00:00' into outfile 'c:/jbpm_log.sql'; #Assume we create tmp table tmp_log create table TMP_LOG (ID_ bigint not null auto_increment, CLASS_ char(1) not null, INDEX_ integer, DATE_ datetime, TOKEN_ bigint, PARENT_ bigint, MESSAGE_ text, EXCEPTION_ text, ACTION_ bigint, NODE_ bigint, ENTER_ datetime, LEAVE_ datetime, DURATION_ bigint, NEWLONGVALUE_ bigint, TRANSITION_ bigint, CHILD_ bigint, SOURCENODE_ bigint, DESTINATIONNODE_ bigint, VARIABLEINSTANCE_ bigint, OLDBYTEARRAY_ bigint, NEWBYTEARRAY_ bigint, OLDDATEVALUE_ datetime, NEWDATEVALUE_ datetime, OLDDOUBLEVALUE_ double precision, NEWDOUBLEVALUE_ double precision, OLDLONGIDCLASS_ varchar(255), OLDLONGIDVALUE_ bigint, NEWLONGIDCLASS_ varchar(255), NEWLONGIDVALUE_ bigint, OLDSTRINGIDCLASS_ varchar(255), OLDSTRINGIDVALUE_ varchar(255), NEWSTRINGIDCLASS_ varchar(255), NEWSTRINGIDVALUE_ varchar(255), OLDLONGVALUE_ bigint, OLDSTRINGVALUE_ text, NEWSTRINGVALUE_ text, TASKINSTANCE_ bigint, TASKACTORID_ varchar(255), TASKOLDACTORID_ varchar(255), SWIMLANEINSTANCE_ bigint, primary key (ID_)) type=InnoDB; load data infile "c:/jbpm_log.sql" into table TMP_LOG; #select * from `jbpm_bytearray` ba where ba.`ID_` IN ( #select vi.`BYTEARRAYVALUE_` from `tmp_variableinstance` vi #) #or ba.`ID_` IN( #select lo.`NEWBYTEARRAY_` from `tmp_log` lo #) #or ba.`ID_` IN ( #select lo.`OLDBYTEARRAY_` from `tmp_log` lo #) into outfile 'c:/jbpm_bytearray0.sql'; #select * from `jbpm_bytearray` ba,`tmp_variableinstance` vi,`tmp_log` lo #where ba.`ID_`=vi.`BYTEARRAYVALUE_` or ba.`ID_`=lo.`NEWBYTEARRAY_` or #ba.`ID_`=lo.`OLDBYTEARRAY_` into outfile 'c:/jbpm_bytearray.sql'; select distinct ID,null,null from ( select `BYTEARRAYVALUE_` AS ID from `tmp_variableinstance` where BYTEARRAYVALUE_ is not null UNION select `NEWBYTEARRAY_` AS ID from `tmp_log` where NEWBYTEARRAY_ is not null UNION select `OLDBYTEARRAY_` AS ID from `tmp_log` where OLDBYTEARRAY_ is not null ) tmp into outfile 'c:/jbpm_bytearray.sql'; create table TMP_BYTEARRAY (ID_ bigint not null auto_increment, NAME_ varchar(255), FILEDEFINITION_ bigint, primary key (ID_)) type=InnoDB; load data infile "c:/jbpm_bytearray.sql" into table TMP_BYTEARRAY; #select * from `jbpm_byteblock` bb where bb.`PROCESSFILE_` IN( #select ID_ from tmp_bytearray #) into outfile 'c:/jbpm_byteblock.sql'; select * from `jbpm_byteblock` bb where bb.`PROCESSFILE_` >= (select min(ID_) from tmp_bytearray) and bb.`PROCESSFILE_` IN( select ID_ from tmp_bytearray ) into outfile 'c:/jbpm_byteblock.sql';

导出数据再导入新数据库再加外键, 加外键是恐怖的了, 暂时在自己的数据上是没问题.

SQL语句写成那样都是有些原因的, 有空再扯...

==================================

产品数据库碰到variableinstance,log都很难导出,故更新

 select * from `jbpm_variableinstance` vi where vi.`PROCESSINSTANCE_` >= (select min(ID_) from tmp_processinstance)
 into outfile 'c:/jbpm_variableinstance.sql';

select * from `jbpm_log` lo where ID_>=26000000 and TO_DAYS(NOW())-TO_DAYS(lo.`DATE_`) <= 45 into outfile 'c:/jbpm_log.sql';

log的ID靠猜出来的, select ID_,DATE_ from jbpm_log, 因为ID_递增, 可以过滤减少查询结果集.

创表加外键之后

SET FOREIGN_KEY_CHECKS=0;

load data infile ........

SET FOREIGN_KEY_CHECKS=1;