诊断案例:从实例挂起到归档失败和内存管理的蝴蝶效应

杨廷琨(yangtingkun)

云和恩墨 CTO

高级咨询顾问,Oracle ACE 总监,ITPUB Oracle 数据库管理版版主

编辑手记:在很多数据库的故障案例中,一个简单的疏忽可能导致问题被层层放大,最终导致故障,这就是蝴蝶效应的传播原理。这里分享的小案例自顶向下的追溯可以显见:实例挂起->归档失败->实例错误->参数配置。根本的原因往往很简单,DBA的严谨尤其重要。

客户的11.2.0.3 RAC数据库出现了归档失败的情况,导致单个实例出现HANG死的状况。

检查错误信息发现:

Tue Jul 02 16:49:13 2013 ARC1: Error 19504 Creating archive log file TO '+DATA02' ARCH: Archival stopped, error occurred. Will continue retrying ORACLE Instance orcl1 - Archival Error ORA-16038: log 14 SEQUENCE# 68244 cannot be archived ORA-19504: failed TO CREATE file "" ORA-00312: online log 14 thread 1: '+DATA02/orcl/onlinelog/group_14.264.792274883' ORA-00312: online log 14 thread 1: '+DATA02/orcl/onlinelog/group_14.265.792274889' Archiver process freed FROM errors. No longer stopped Tue Jul 02 16:50:37 2013 ARC0: LGWR IS actively archiving destination LOG_ARCHIVE_DEST_3 ARCH: Archival stopped, error occurred. Will continue retrying ORACLE Instance orcl1 - Archival Error ORA-16014: log 14 SEQUENCE# 68244 NOT archived, no available destinations ORA-00312: online log 14 thread 1: '+DATA02/orcl/onlinelog/group_14.264.792274883' ORA-00312: online log 14 thread 1: '+DATA02/orcl/onlinelog/group_14.265.792274889' ARC0: Archive log rejected (thread 1 SEQUENCE 68240) at host 'orclsh' FAL[server, ARC0]: FAL archive failed, see trace file. ARCH: FAL archive failed. Archiver continuing ORACLE Instance orcl1 - Archival Error. Archiver continuing.

由于归档失败发生在ASM磁盘上,首先检查ASM磁盘空间以及DB_RECOVERY_FILE_DEST_SIZE,ASM磁盘空间是足够的,而且由于只有一个节点出现出现了无法归档的问题,也可以排除是空间不足造成的。

确认两个节点的DB_RECOVERY_FILE_DEST_SIZE参数设置都是0,基本上可以判断问题和当前节点的ASM实例状态不正常有关。

接下来检查ASM实例的错误信息:

Tue Jul 02 16:41:43 2013 Dumping diagnostic DATA IN directory=[cdmp_20130702164115], requested BY (instance=2, osid=2032294 (LMD0)), summary=[incident=165521]. Tue Jul 02 16:49:13 2013 Dumping diagnostic DATA IN directory=[cdmp_20130702164845], requested BY (instance=2, osid=2032294 (LMD0)), summary=[incident=165522].

当前节点ASM实例出现了的这个信息,说明报错发生在实例2上,检查实例2的ASM告警日志文件:

Tue Jul 02 18:29:55 2013 Errors IN file /u01/app/grid/diag/asm/+asm/+ASM2/trace/+ASM2_lmd0_2032294.trc (incident=186256): ORA-04031: unable TO allocate 3768 bytes OF shared memory ("shared pool","unknown object","sga heap(1,0)","ges enqueues") USE ADRCI OR Support Workbench TO package the incident. See Note 411.1 at My Oracle Support FOR error AND packaging details. Insufficient shared pool TO allocate a GES object (ospid 2032294) Tue Jul 02 18:29:55 2013 Sweep [inc][186256]: completed Tue Jul 02 18:36:49 2013 Errors IN file /u01/app/grid/diag/asm/+asm/+ASM2/trace/+ASM2_lmd0_2032294.trc (incident=186257): ORA-04031: unable TO allocate 3768 bytes OF shared memory ("shared pool","unknown object","sga heap(1,0)","ges enqueues") USE ADRCI OR Support Workbench TO package the incident. See Note 411.1 at My Oracle Support FOR error AND packaging details. Insufficient shared pool TO allocate a GES object (ospid 2032294)

果然实例2上的ASM出现了大量ORA-4031错误,这说明ASM实例的Shared Pool内存可能配置不足。检查ASM启动的参数配置:

Sat Aug 25 20:06:55 2012 NOTE: No asm libraries found IN the system ERROR: -5(Duplicate disk DATA_DG01:ASM_DISK1) ERROR: -5(Duplicate disk DATA_DG01:ASM_DISK2) MEMORY_TARGET defaulting TO 411041792. * instance_number obtained FROM CSS = 2, checking FOR the existence OF node 0... * node 0 does NOT exist. instance_number = 2 Starting ORACLE instance (normal) LICENSE_MAX_SESSION = 0 LICENSE_SESSIONS_WARNING = 0 Private Interface 'en1' configured FROM GPnP FOR USE AS a private interconnect. [name='en1', TYPE=1, ip=169.254.78.6, mac=00-1a-64-bb-50-7d, net=169.254.0.0/16, mask=255.255.0.0, USE=haip:cluster_interconnect/62] Public Interface 'en0' configured FROM GPnP FOR USE AS a public interface. [name='en0', TYPE=1, ip=10.1.16.35, mac=00-1a-64-bb-50-7c, net=10.1.16.32/27, mask=255.255.255.224, USE=public/1] Picked latch-free SCN scheme 3 USING LOG_ARCHIVE_DEST_1 parameter DEFAULT VALUE AS /u01/app/11.2.0.3/grid/dbs/arch Autotune OF undo retention IS turned ON. LICENSE_MAX_USERS = 0 SYS auditing IS disabled NOTE: Volume support enabled Starting up: Oracle DATABASE 11g Enterprise Edition Release 11.2.0.3.0 - 64bit Production WITH the REAL Application Clusters AND Automatic Storage Management options. ORACLE_HOME = /u01/app/11.2.0.3/grid System name: AIX Node name: orcldb2 Release: 1 Version: 6 Machine: 00C94E064C00 USING parameter settings IN server-side pfile /u01/app/11.2.0.3/grid/dbs/init+ASM2.ora System parameters WITH non-DEFAULT VALUES: large_pool_size = 12M instance_type = "asm" remote_login_passwordfile= "EXCLUSIVE" asm_diskstring = "/dev/ocr_*" asm_diskstring = "/dev/voting_*" asm_diskstring = "/dev/asm_*" asm_diskgroups = "DATA" asm_diskgroups = "DATA_DG01" asm_diskgroups = "SPFILE_DG" asm_power_limit = 1 diagnostic_dest = "/u01/app/grid" Cluster communication IS configured TO USE the following interface(s) FOR this instance 169.254.78.6 cluster interconnect IPC version:Oracle UDP/IP (generic) IPC Vendor 1 proto 2

当前ASM实例使用默认的MEMORY_TARGET配置,分配大小大约是400M,根据Oracle的MOS文章:ASM & Shared Pool (ORA-4031) [ID 437924.1],在11.2.0.3中,Oracle增加了ASM实例所允许的默认进程数PROCESSES,但是默认的MEMORY_TARGET参数没有增加。

根据Oracle的建议,11.2.0.3的MEMORY_TARGET至少应该设置到1536M,而MEMORY_MAX_TARGET设置为4096M

SQL> ALTER system SET memory_max_target=4096m scope=spfile; SQL> ALTER system SET memory_target=1536m scope=spfile;

对于当前的情况,如果短时间内无法重启DB和ASM实例,可以在问题节点配置一个第二本地归档路径,设置目标路径为本地磁盘,从而避免归档无法完成而导致的实例HANG死。

原文发布于微信公众号 - 数据和云(OraNews)

原文发表时间:2016-05-30

本文参与腾讯云自媒体分享计划,欢迎正在阅读的你也加入,一起分享。

发表于

我来说两句

0 条评论
登录 后参与评论

相关文章

来自专栏数据库新发现

Use Nid to Change dbname

http://www.eygle.com/faq/Use.Nid.to.Change.Your.dbname.htm

673
来自专栏杨建荣的学习笔记

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。 ...

3254
来自专栏醉梦轩

Linux上C语言写的简易telnet客户端

972
来自专栏杨建荣的学习笔记

redo日志文件学习(22天)

关于redo日志文件,今天因为要做redo日志的在线迁移,所以特意做了个简单的总结。 1. 如果要把redo的切换信息显示到alert日志中,需要设置个参数。...

3315
来自专栏杨建荣的学习笔记

ORA-01113问题的简单分析(r6笔记第3天)

在启动数据库的时候,open阶段总是可能出现各种各样的问题, 比如让人胆战心惊的错误。 ORA-01113: file 1 needs media recov...

2355
来自专栏杨建荣的学习笔记

手工创建/删除数据库的步骤

今天和大家分享下数据库的创建和删除的步骤,里面有很多细节需要大家考虑。创建数据库不只是一个create database语句。删除数据库 drop databa...

3016
来自专栏杨建荣的学习笔记

impdp异常中断导致的问题(r2第8天)

今天查看数据库的负载,发现cpu消耗异常的高。里面有不少dw的进程.但是查看impdp的进程却不存在。 查看datapump的进程情况,发现大量的job,但是状...

35811
来自专栏杨建荣的学习笔记

关于数据库中的一些name(r3笔记第64天)

如果接触数据库有些时间了,可能会碰到很多关于数据库相关的名字,比如ORACLE_SID,db_name,instance_name,db_unique_name...

31812
来自专栏数据和云

偷梁换柱 | 无备份情况下的数据恢复实践

在实际环境中,许多数据库环境并没有做好完整的数据备份恢复计划及容灾方案,无法保证数据安全,并且出现一些灾难性的错误。那么我们就面临这样的问题:在什么样的最极端情...

3025
来自专栏乐沙弥的世界

ORA-12537 TNS-12518 Process m000 died

      最近客户邮件描述无法从客户端连接到数据库,其错误号为ORA-12537: TNS:connection closed,连接被关闭。直接通过tnspi...

793

扫描关注云+社区