案例详解:Linux文件系统异常导致数据库文件无法访问

墨墨导读:某客户单位数据库出现异常,大致现象是:数据库状态是open的,但是其中一个数据文件无法访问,本文分享排查原因与解决问题的整个过程。

通过ls 查看文件都报错,如下所示:

[oracle@oracledata01 oracle]$ ls
ls: 无法访问zf4.dbf: 输入/输出错误


同时dd命令也无法读取该文件:

[oracle@oracledata01 fd]$ dd if=zf4.dbf bs=8192 count=1 skip=1|od -x|header -1
bash: header: command not found
dd: 正在打开"zf4.dbf": 没有那个文件或目录


此时在从数据库层面看到的信息类似如下:

Tue Apr 28 16:49:51 2020
Errors in file /oracle_app/Oracle12c/diag/rdbms/orcl/orcl/trace/orcl_ora_9146.trc:
ORA-01157: 无法标识/锁定数据文件 11 - 请参阅 DBWR 跟踪文件
ORA-01110: 数据文件 11: '/oracle_data/oracle/zf4.dbf'
ORA-1157 signalled during: ALTER DATABASE OPEN...
Tue Apr 28 16:49:51 2020
Errors in file /oracle_app/Oracle12c/diag/rdbms/orcl/orcl/trace/orcl_m000_9152.trc:
ORA-27037: 无法获得文件状态
Linux-x86_64 Error: 5: Input/output error
Additional information: 3
ORA-01122: 数据库文件 11 验证失败
ORA-01110: 数据文件 11: '/oracle_data/oracle/zf4.dbf'
ORA-01565: 标识文件 '/oracle_data/oracle/zf4.dbf' 时出错
ORA-27037: 无法获得文件状态
Linux-x86_64 Error: 5: Input/output error
Additional information: 3
Tue Apr 28 16:50:19 2020
USER (ospid: 9309): terminating the instance


从上述信息来看,数据库第11号文件无法读取。

在我们进行恢复之前,客户进行了多次尝试,还遭遇了一些数据库坏块,也遇到了不少ora-00600 错误:

Wed Apr 29 03:10:12 2020
Errors in file /oracle_app/Oracle12c/diag/rdbms/orcl/orcl/trace/orcl_j009_65816.trc:
ORA-12012: 自动执行作业 "ZX_XT"."JS_XT_XZGLXX_XZXK_FR" 出错
ORA-06502: PL/SQL: 数字或值错误 :  字符串缓冲区太小
ORA-06512: 在 "ZX_XT.PKG_双公示接收_邢_2019", line 1071
ORA-00942: 表或视图不存在
ORA-02063: 紧接着 line (起自 BM_SJJH)
alter database datafile 11 resize 20g
Completed: alter database datafile 11 resize 20g
Wed Apr 29 03:10:15 2020
Corrupt Block Found
CONT = 0, TSN = 1, TSNAME = SYSAUX
RFN = 3, BLK = 3, RDBA = 12582915
OBJN = 1, OBJD = -1, OBJECT = _NEXT_OBJECT, SUBOBJECT =
SEGMENT OWNER = SYS, SEGMENT TYPE = Invalid Type
Wed Apr 29 03:10:15 2020
Errors in file /oracle_app/Oracle12c/diag/rdbms/orcl/orcl/trace/orcl_m006_66106.trc  (incident=246250):
ORA-00600: 内部错误代码, 参数: [kdBlkCheckError], [3], [3], [18018], [], [], [], [], [], [], [], []
Incident details in: /oracle_app/Oracle12c/diag/rdbms/orcl/orcl/incident/incdir_246250/orcl_m006_66106_i246250.trc
Wed Apr 29 03:10:15 2020
......
Wed Apr 29 03:11:00 2020
Doing block recovery for file 3 block 3
Resuming block recovery (PMON) for file 3 block 3
Block recovery from logseq 50078, block 15827 to scn 14514058257813
Wed Apr 29 03:11:00 2020
Recovery of Online Redo Log: Thread 1 Group 3 Seq 50078 Reading mem 0
Mem# 0: /oracle_data/oradata/orcl/redo03.log
Block recovery completed at rba 50078.23957.16, scn 3379.1363764630
Errors in file /oracle_app/Oracle12c/diag/rdbms/orcl/orcl/trace/orcl_m004_66271.trc  (incident=245043):
ORA-00600: 内部错误代码, 参数: [kdBlkCheckError], [3], [3], [18018], [], [], [], [], [], [], [], []
Incident details in: /oracle_app/Oracle12c/diag/rdbms/orcl/orcl/incident/incdir_245043/orcl_m004_66271_i245043.trc
Use ADRCI or Support Workbench to package the incident.
See Note 411.1 at My Oracle Support for error and packaging details.
Wed Apr 29 03:11:01 2020
Errors in file /oracle_app/Oracle12c/diag/rdbms/orcl/orcl/trace/orcl_m004_66271.trc:
ORA-00600: 内部错误代码, 参数: [kdBlkCheckError], [3], [3], [18018], [], [], [], [], [], [], [], []
Wed Apr 29 03:11:02 2020
......
Wed Apr 29 03:15:49 2020
Errors in file /oracle_app/Oracle12c/diag/rdbms/orcl/orcl/trace/orcl_j000_65779.trc  (incident=245026):
ORA-00600: internal error code, arguments: [13011], [7907], [12713924], [44], [12687314], [0], [], [], [], [], [], []
ORA-01403: no data found
ORA-06512: at "SYS.DBMS_AQ_INV", line 1248
ORA-06512: at line 1
Incident details in: /oracle_app/Oracle12c/diag/rdbms/orcl/orcl/incident/incdir_245026/orcl_j000_65779_i245026.trc
Wed Apr 29 03:15:52 2020
Dumping diagnostic data in directory=[cdmp_20200429031552], requested by (instance=1, osid=65779 (J000)), summary=[incident=245026].
Use ADRCI or Support Workbench to package the incident.
See Note 411.1 at My Oracle Support for error and packaging details.
Errors in file /oracle_app/Oracle12c/diag/rdbms/orcl/orcl/trace/orcl_j000_65779.trc  (incident=245027):
ORA-00600: internal error code, arguments: [13011], [7907], [12713924], [44], [12687314], [0], [], [], [], [], [], []
ORA-01403: no data found
ORA-06512: at "SYS.DBMS_AQ_INV", line 1248
ORA-06512: at line 1
Incident details in: /oracle_app/Oracle12c/diag/rdbms/orcl/orcl/incident/incdir_245027/orcl_j000_65779_i245027.trc
Use ADRCI or Support Workbench to package the incident.
......
Wed Apr 29 04:05:07 2020
Errors in file /oracle_app/Oracle12c/diag/rdbms/orcl/orcl/trace/orcl_m003_76294.trc:
ORA-00600: 内部错误代码, 参数: [kdBlkCheckError], [3], [3], [18018], [], [], [], [], [], [], [], []
Wed Apr 29 04:05:08 2020
Dumping diagnostic data in directory=[cdmp_20200429040508], requested by (instance=1, osid=76263), summary=[incident=282250].
Errors in file /oracle_app/Oracle12c/diag/rdbms/orcl/orcl/trace/orcl_ora_76263.trc  (incident=282251):
ORA-00600: 内部错误代码, 参数: [qosdFindObjRead: objcnt mismatch], [], [], [], [], [], [], [], [], [], [], []
Incident details in: /oracle_app/Oracle12c/diag/rdbms/orcl/orcl/incident/incdir_282251/orcl_ora_76263_i282251.trc
Use ADRCI or Support Workbench to package the incident.
See Note 411.1 at My Oracle Support for error and packaging details.
Wed Apr 29 04:05:10 2020
Sweep [inc][282251]: completed
Sweep [inc2][282265]: completed
Sweep [inc2][282251]: completed
Wed Apr 29 04:05:10 2020
Dumping diagnostic data in directory=[cdmp_20200429040510], requested by (instance=1, osid=76263), summary=[incident=282251].
Errors in file /oracle_app/Oracle12c/diag/rdbms/orcl/orcl/trace/orcl_ora_76263.trc  (incident=282252):
ORA-00600: 内部错误代码, 参数: [qosdFindObjRead: objcnt mismatch], [], [], [], [], [], [], [], [], [], [], []
Incident details in: /oracle_app/Oracle12c/diag/rdbms/orcl/orcl/incident/incdir_282252/orcl_ora_76263_i282252.trc
Use ADRCI or Support Workbench to package the incident.
See Note 411.1 at My Oracle Support for error and packaging details.
Wed Apr 29 04:07:26 2020
......
Wed Apr 29 04:24:01 2020
Sweep [inc][279857]: completed
Sweep [inc2][279857]: completed
Errors in file /oracle_app/Oracle12c/diag/rdbms/orcl/orcl/trace/orcl_ora_78503.trc  (incident=279858):
ORA-00600: 内部错误代码, 参数: [qosdFindObjRead: objcnt mismatch], [], [], [], [], [], [], [], [], [], [], []
Incident details in: /oracle_app/Oracle12c/diag/rdbms/orcl/orcl/incident/incdir_279858/orcl_ora_78503_i279858.trc
Use ADRCI or Support Workbench to package the incident.
See Note 411.1 at My Oracle Support for error and packaging details.
Wed Apr 29 04:24:03 2020
Sweep [inc][279858]: completed
Sweep [inc2][279858]: completed
Wed Apr 29 04:24:04 2020
Dumping diagnostic data in directory=[cdmp_20200429042404], requested by (instance=1, osid=78503), summary=[incident=279858].
Wed Apr 29 04:25:57 2020
Errors in file /oracle_app/Oracle12c/diag/rdbms/orcl/orcl/trace/orcl_ora_78738.trc  (incident=282222):
ORA-00600: 内部错误代码, 参数: [qosdFindObjRead: objcnt mismatch], [], [], [], [], [], [], [], [], [], [], []
Incident details in: /oracle_app/Oracle12c/diag/rdbms/orcl/orcl/incident/incdir_282222/orcl_ora_78738_i282222.trc
Use ADRCI or Support Workbench to package the incident.
See Note 411.1 at My Oracle Support for error and packaging details.
Wed Apr 29 04:25:58 2020
Sweep [inc][282222]: completed
Sweep [inc2][282222]: completed
Wed Apr 29 04:25:58 2020
Dumping diagnostic data in directory=[cdmp_20200429042558], requested by (instance=1, osid=78738), summary=[incident=282222].
Errors in file /oracle_app/Oracle12c/diag/rdbms/orcl/orcl/trace/orcl_ora_78738.trc  (incident=282223):
ORA-00600: 内部错误代码, 参数: [qosdFindObjRead: objcnt mismatch], [], [], [], [], [], [], [], [], [], [], []
Incident details in: /oracle_app/Oracle12c/diag/rdbms/orcl/orcl/incident/incdir_282223/orcl_ora_78738_i282223.trc


上述ora-00600 错误,总的来都是常见的,几乎都可以定性为数据坏块。

对于file 3的坏块,处理相对简单,直接drop或者truncate相关对象即可,然后重建awr。

但是对于第11号数据文件,怎么办呢?这里我们的处理方法是当文件丢失处理(很久之前处理过Windows环境数据文件大小为 0 kb的问题,这几种情况都类似。

这里我说一下简单的处理思路:

1. 通过odu 扫盘,分别以不同offset方式来扫(0和4096)。

2. 对比两次扫描结果,组合extent信息,抽取文件。

由于这里文件只有1个,大小20g左右,通过dd 拼接即可。

最终文件拼接完成后,由于这里是非归档环境,无法进行正常recover,因此还需要通过bbed来修改数据文件头的checkpoint信息。

对于文件系统来讲,目前已经不是Oracle存储的最佳选择了,建议使用ASM。当然,其他数据库比如MySQL、PostgreSQL除外。

推荐阅读:144页!分享珍藏已久的数据库技术年刊

数据和云

ID:OraNews

如有收获,请划至底部,点击“在看”,谢谢!

点击下图查看更多 ↓

云和恩墨大讲堂 | 一个分享交流的地方

长按,识别二维码,加入万人交流社群

请备注:云和恩墨大讲堂

  点个“在看”

你的喜欢会被看到❤

原文地址:https://www.cnblogs.com/hzcya1995/p/13311670.html