一条sql语句“导致”的数据库宕机问题及分析 （38天)

最近测试环境需要做一些变更，把测试环境切分成两套环境，存储空间也需要压缩压缩和整理。 unix组的人已经开始做空间划分了，然后我们需要在此基础上重建一套环境。 有些数据文件使用空间不大，所以准备压缩一下。 用了下面的sql语句，结果跑了十几秒中就抛了下面的错误。

SQL> set linesize 200
SQL> col name for a40
SQL> col resizecmd for a80
SQL> select a.file#,a.name,a.bytes/1024/1024 CurrentMB,
  2         ceil(HWM * a.block_size)/1024/1024 ResizeTo,
  3         (a.bytes - HWM * a.block_size)/1024/1024 ReleaseMB,
  4         'alter database datafile '''||a.name||''' resize '||
  5         ceil(HWM * a.block_size/1024/1024) || 'M;' ResizeCMD
  6  from v$datafile a,
  7       (select file_id,max(block_id+blocks-1) HWM
  8         from dba_extents
  9         group by file_id) b
 1	0  where a.file# = b.file_id(+)
 11  and (a.bytes - HWM *block_size)>0
order by 5     12  ;
order by 5
             *
ERROR at line 12:
ORA-01116: error in opening database file 4
ORA-01110: data file 4: '/testPT1/oracle/TEST01/oratmp01/temp/ggs_data01.dbf'
ORA-27041: unable to open file
Linux-x86_64 Error: 2: No such file or directory
Additional information: 3

再一查看数据库进程，都没有了，看来数据库是宕了。 我还想这条sql语句真是厉害，看看日志里面怎么说。 Tue Mar 25 22:04:19 2014 End automatic SQL Tuning Advisor run for special tuning task "SYS_AUTO_SQL_TUNING_TASK" Wed Mar 26 02:00:00 2014 Closing scheduler window Closing Resource Manager plan via scheduler window Clearing Resource Manager plan via parameter Mon Mar 31 10:14:55 2014 USER (ospid: 21846): terminating the instance due to error 472 Instance terminated by USER, pid = 21846 先把库重启了，看有没有什么问题。

SQL> startup nomount ORA-32004: obsolete or deprecated parameter(s) specified for RDBMS instance ORACLE instance started. Total System Global Area 1.2827E+10 bytes Fixed Size 2253880 bytes Variable Size 4211084232 bytes Database Buffers 8589934592 bytes Redo Buffers 24096768 bytes SQL> alter database mount; Database altered. 当准备Open的时候，报了下面的错误。说有个数据文件丢失了。 SQL> alter database open; alter database open * ERROR at line 1: ORA-01157: cannot identify/lock data file 4 - see DBWR trace file ORA-01110: data file 4: '/testPT1/oracle/TEST01/oratmp01/temp/ggs_data01.dbf' 查看备份情况，是否有热备份之类的操作遗留。可以看到这个文件确实是损坏了或者被认为删除了。 SQL> select *from v$backup; FILE# STATUS CHANGE# TIME ---------- ------------------ ---------- --------- 1 NOT ACTIVE 1.0583E+13 22-JAN-13 2 NOT ACTIVE 1.0583E+13 22-JAN-13 3 NOT ACTIVE 1.0583E+13 22-JAN-13 4 FILE NOT FOUND 0 ... 40 NOT ACTIVE 1.0583E+13 22-JAN-13 查看文件的路径。 1* select file#,name from v$datafile 4 /testPT1/oracle/TEST01/oratmp01/temp/ggs_data01.dbf 确认了下这个数据文件goldengate用，现在测试环境上还没有goldengate，可以删除，于是头脑一发热，就准备马上删了，结果报了错。 SQL> drop tablespace GGS_DATA including contents and datafiles cascade constraint; drop tablespace GGS_DATA including contents and datafiles cascade constraint * ERROR at line 1: ORA-01109: database not open 才反应过来数据库还在mount状态 先把数据文件offline了 SQL> alter database datafile '/testPT1/oracle/TEST01/oratmp01/temp/ggs_data01.dbf' offline; Database altered. 这时候再查看v$backup,那条记录就不复存在了。 FILE# STATUS CHANGE# TIME ---------- ------------------ ---------- --------- 1 NOT ACTIVE 1.0583E+13 22-JAN-13 2 NOT ACTIVE 1.0583E+13 22-JAN-13 3 NOT ACTIVE 1.0583E+13 22-JAN-13 5 NOT ACTIVE 1.0583E+13 22-JAN-13 6 NOT ACTIVE 1.0583E+13 22-JAN-13 7 NOT ACTIVE 1.0583E+13 22-JAN-13 8 NOT ACTIVE 1.0583E+13 22-JAN-13 9 NOT ACTIVE 1.0583E+13 22-JAN-13 10 NOT ACTIVE 1.0583E+13 22-JAN-13 11 NOT ACTIVE 1.0583E+13 22-JAN-13 12 NOT ACTIVE 1.0583E+13 22-JAN-13 .... 39 rows selected. 把数据库open起来。 SQL> alter database open; Database altered. 然后再删除，其实这个阶段也只是释放了句柄和更新了数据字典。 SQL> drop tablespace ggs_data including contents and datafiles cascade constraint; Tablespace dropped. 再次查看数据库进程是否正常。可以看到数据库已经正常了。 test01@ccbdbpt4:/opt/app/oracle/TEST01> ps -ef|grep smon test018420 5954 0 10:39 pts/2 00:00:00 grep smon test0110295 1 0 10:18 ? 00:00:01 ora_smon_TEST01 test01@ccbdbpt4:/opt/app/oracle/TEST01> sqlplus / as sysdba SQL*Plus: Release 11.2.0.2.0 Production on Mon Mar 31 10:40:08 2014 Copyright (c) 1982, 2010, Oracle. All rights reserved. Connected to: Oracle Database 11g Enterprise Edition Release 11.2.0.2.0 - 64bit Production With the Partitioning, OLAP, Data Mining and Real Application Testing options SQL> show user USER is "SYS" SQL> select status from v$instance; STATUS ------------ OPEN 我在另一个环境上碰到了类似的问题。 test01@ccbdbpt4:/opt/app/oracle/TEST01/test> sqlplus / as sysdba SQL*Plus: Release 11.2.0.2.0 Production on Thu Apr 3 17:26:43 2014 Copyright (c) 1982, 2010, Oracle. All rights reserved. Connected to: Oracle Database 11g Enterprise Edition Release 11.2.0.2.0 - 64bit Production With the Partitioning, OLAP, Data Mining and Real Application Testing options SQL> SQL> select dbms_metadata.get_ddl('USER',u.username) from dba_users u WHERE USER NAME in('TEST'); ERROR: ORA-03113: end-of-file on communication channel Process ID: 4056 Session ID: 4159 Serial number: 15 no rows selected 查看日志，错误还是类似。 Mon Mar 31 11:47:47 2014 USER (ospid: 26945): terminating the instance due to error 472 Instance terminated by USER, pid = 26945 这个问题的总结如下 导致数据库宕机的原因不是因为sql语句，而是因为unix组做文件的操作中，导致数据库进程问题，数据库其实已经停了，但是我们仍然可以进行简单的操作。因为操作系统中句柄还在。 那个数据文件的丢失和unix组的人确认是认为的失误，他们在做空间切分的时候，没有停库。