123
 123

Tip: 看不到本站引用 Flickr 的图片? 下载 Firefox Access Flickr 插件 | AD: 订阅 DBA notes --

2010-06-04 Fri

19:46 恩墨科技备份恢复培训课程成功举行中 (3388 Bytes) » Oracle Life

作者:eygle 发布在 eygle.com

这一周是繁忙的一周,恩墨科技的备份恢复课程也成功举行,现在课程正在进行中。李轶楠同学已经完成了他的两天课程,今明两天是崔华的主场,我在现场来学习崔华的经验。

课程的环境和氛围比我想象的还要好,培训是一个现场互动性非常强的工作,我建议大家可以把培训看成是一次座谈,大家可以充分和讲师互动来促进理解,建议讲师用更合适的方法来进行授课。
我们双方的目标是一致的:通过现场互动培训,将有用的知识以易于理解的方式传达给大家

600前两天做了非常完善的讲解与铺垫,以实践可用的操作为主,将日常工作必要的技能传授出来。
而崔华讲解的将是更加深入的内容,从本质上介绍Oracle的恢复机制。
我戏称,要控制崔华,不要讲的太深、太抽象,他现场控制的非常好,而我也从他这里学到好多东西,取长补短,我相信我以后在备份恢复方面一定可以传达出更好的内容。

而Oracle的确是博大精深,我觉得关于Redo方面的内容,完全都可以组织一个非常精彩的,至少2天的课程。

有学员问到,你们如何知道这样的方法,这样的手段?
我说,我们真的是花了5、6年才能理解到这个层面,而通过培训,现在大家可以快速的获得这些知识,这就是进步。而我非常高兴我们能够通过培训、甚至是ACOUG的公开免费活动将知识和经验传播出去。

感谢朋友们的支持!


相关文章|Related Articles

评论数量(0)|Add Comments

本文网址:

13:54 Log Buffer #191, A Carnival of the Vanities for DBAs (4862 Bytes) » The Pythian Blog

Welcome to Log Buffer, the weekly roundup of database blogs.

Kicking off this week in Log Buffer #191 are posts from Alisher Yuldashev:

Randolf Geist blogs on an Advanced Oracle Troubleshooting Session – PGA/UGA memory fragmentation for when a batch process takes significantly longer than expected.

James Morle talks about an example of a misleading average in Log File Sync and AWR – Not Good Bedfellows.

And a few faves from Bradd Piontek:

Marco Gralike, on Blog.Gralike.Com, revisits Enabling and Disabling Database Options, a small item that is easily overlooked. Marco also notes a cool tool: VirtualBox Appliance which makes a great start-up test environment. Word of caution however, it’s for testing purposes only.

On Askdba.org, Amit advises on downloading Oracle software directly to server in a post based on Pythian’s downloading from OTN directly to your database server. Watch for future posts from Brad on how he does it via Firefox, and edelivery.oracle.com.

Alex Gorbachev is spreading the word about The Ultimate SQL Tune-off with Jonathan Lewis and Kyle Hailey, two of his most respected Oracle performance experts, believing the session should be interesting to all DBAs, not just Oracle.

Robert Catteral continues to recap session highlights from the International DB2 Users Group Conference last month in Nuggets from DB2 by the Bay, Part 3, following Parts 1 & 2.

Chen Shapira contributed Cloning Oracle Home from RAC to Stand-Alone.

On In Recovery, Paul Randal wrote the whitepaper Proven SQL Server Architectures for High Availability and Disaster Recovery he wrote for the Spring SQL Server release has been published.

Moving to MySQL world, Vadim Tkachenko continues storage benchmarking of MySQL FlashCache (very much like Oracle FlashCache but for MySQL InnoDB engine). This time he is using FusionIO cards for FlashCache.

And, to round things off, Ronald Bradford writes about When SET GLOBAL affects SESSION scope.

Have a great weekend everyone.

12:48 Quiz Night (1 Bytes) » Oracle Scratchpad
08:32 Ultimate SQL Tune-off with Jonathan Lewis and Kyle Hailey (2067 Bytes) » The Pythian Blog

This is just a short blog post to spread the word about the upcoming webinar organized by Embarcadero with fellow member of OakTable Network — Jonathan Lewis and Kyle HaileyUltimate SQL Tune-off.

Ultimate SQL Tune-off

Kyle and Jonathan are on my list of the most respected Oracle performance experts and I could only imagine what they can deliver joining the forces. I’ll allow myself to quote the opening of the webinar as it sounds really interesting:

According to Jonathan Lewis, there is a basic strategy to writing efficient SQL, but it requires knowledge of the data, the metadata, and the intent of the query.

Jonathan will start with the basic principles of efficient data access and then show different ways of analyzing the data. He’ll then describe a graphical approach to finding the most efficient access path for a query and then apply the method to a problem he recently saw with one of his clients.

But wait… is there more than one way to skin a cat? Kyle Hailey thinks so, and he’s not afraid to let Jonathan know it!

Kyle’s fancy footwork will include his own approach to Visual SQL Tuning (VST) with some ring side support from DB Optimizer XE. Hitting below the belt? You be the judge.

The webinar is on June 10, 2010 at 11AM Pacific / 2PM Eastern.

06:15 Silent installation of Enteprise Manager Grid Control 11g (7946 Bytes) » The Pythian Blog

One of the first questions that I asked myself when EM GC 11g came out is how would a silent
installation go especially taking into account the use of WebLogic server as the http server.

But before any attempt of silent installation I started the GC 11g GUI installation to local VM machine
(Oracle Enterprise Linux 5.4 64-bit) with an allocated 2Gb of memory and got errors since
WebLogic server was not installed there.

I downloaded it from edelivery.oracle.com and unpacked zip file – got one jar file! Not bad at all!
But where are all those runInstaller, stage, & other directories? There was not even a sample silent
response file. This is where documentation is helpful, and based on Creating a Silent Installation,
I created response file silent.xml

<?xml version="1.0" encoding="UTF-8" ?>
<domain-template-descriptor>
<input-fields>
   <data-value name="BEAHOME" value="/u01/app/oracle/product/gc11" />
   <data-value name="WLS_INSTALL_DIR" value="/u01/app/oracle/product/gc11/wlserver_10.3" />
   <data-value name="COMPONENT_PATHS" value="WebLogic Server" />
</input-fields>
</domain-template-descriptor>

and tried to start installation but got an error that java should be at least 1.6 version.
After installation of JRE out of jre-6u20-linux-x64-rpm.bin I changed the symbolic link
and pointed new version java to /usr/bin/java and completed silent installation of
WebLogic Server 10.3.0.2:

java -jar wls1032_generic.jar -mode=silent -silent_xml=silent.xml -log=install.log
Extracting 0%.............................................................................100%
May 6, 2010 9:33:34 PM java.util.prefs.FileSystemPreferences$2 run
INFO: Created user preferences directory.

Intstalling Grid Control requires running a database where the repository will be created.
Oracle RDBMS 11.2 was installed silently as well and that installation is even more silent
than previous versions since there is no progress bar now for 11.2.


./runInstaller -silent \
-responseFile /home/oracle/install/db112/database/response/db_install.rsp \
oracle.install.option=INSTALL_DB_SWONLY \
oracle.install.db.InstallEdition=EE \
security_updates_via_myoraclesupport=false \
decline_security_updates=true \
oracle.install.db.DBA_GROUP=dba \
oracle.install.db.OPER_GROUP=dba \
ORACLE_BASE="/u01/app/oracle" \
ORACLE_HOME="/u01/app/oracle/product/11.2.0/dbhome_1"
Starting Oracle Universal Installer...

Database emrep was created on file system taking into account database parameter requirements
for the repository database at Setting Database Initialization Parameters

./dbca -silent -createDatabase -gdbName emrep -templateName General_Purpose.dbc \
-emConfiguration none -datafileDestination /u01/oradata \
-sysPassword sys_pwd -systemPassword sys_pwd -storageType FS \
-initParams filesystemio_options=setall,job_queue_processes=10,log_buffer=10485760,\
open_cursors=300,processes=500,session_cached_cursors=200 \
-totalMemory 600

One pre-requisite is to have undo tablespace not less than 200Mb. I increased the size of the datafile, followed with a database control deletion from the created database:

emca -deconfig dbcontrol db -repos drop -silent -SYS_PWD sys_pwd -SYSMAN_PWD sys_pwd -SID emrep -PORT 1521

Before proceeding with further GC silent installation I checked Prerequisites
and recognized that “soft nofile” limits for an oracle user are required to be set to 4096 instead of 1024
recommended for oracle 11.2 RDBMS installation.
After change of kernel parameters I put values to response file for silent installation of Grid Control 11:

RESPONSEFILE_VERSION=2.2.1.0.0
SECURITY_UPDATES_VIA_MYORACLESUPPORT=false
DECLINE_SECURITY_UPDATES=true
INSTALL_UPDATES_SELECTION="skip"
ORACLE_MIDDLEWARE_HOME_LOCATION=/u01/app/oracle/product/gc11
ORACLE_INSTANCE_HOME_LOCATION=/u01/app/oracle/product/gc11/gc_inst
WLS_ADMIN_SERVER_USERNAME=weblogic
WLS_ADMIN_SERVER_PASSWORD=Mon10day10
WLS_ADMIN_SERVER_CONFIRM_PASSWORD=Mon10day10
NODE_MANAGER_PASSWORD=Mon10day10
NODE_MANAGER_CONFIRM_PASSWORD=Mon10day10
DATABASE_HOSTNAME="oel7.pythian"
LISTENER_PORT="1521"
SERVICENAME_OR_SID="emrep"
SYS_PASSWORD="sys_pwd"
MANAGEMENT_TABLESPACE_LOCATION="/u01/oradata/emrep/mgmt.dbf"
CONFIGURATION_DATA_TABLESPACE_LOCATION="/u01/oradata/emrep/mgmt_ecm_depot1.dbf"
JVM_DIAGNOSTICS_TABLESPACE_LOCATION="/u01/oradata/emrep/mgmt_ad4j.dbf"
START_OMS=true
SYSMAN_PASSWORD=Mon10day10
SYSMAN_CONFIRM_PASSWORD=Mon10day10
ALLOW_ONLY_SECURE_ACCESS_TO_CONSOLE=true
AGENT_REGISTRATION_PASSWORD=Mon10day10
AGENT_REGISTRATION_CONFIRM_PASSWORD=Mon10day10
FROM_LOCATION="../oms/Disk1/stage/products.xml"
DEINSTALL_LIST={"oracle.sysman.top.oms","11.1.0.1.0"}
TOPLEVEL_COMPONENT={"oracle.sysman.top.oms","11.1.0.1.0"}
COMPONENT_LANGUAGES={"en"}
b_upgrade=false
EM_INSTALL_TYPE="NOSEED"

Installation ran almost all the way through but failed at OMS Configuration Assistant step:

Running Configuration assistant "OMS Configuration"
Executing the OMSCA command...
Check the OMS Configuration Assistant logs at: /u01/app/oracle/product/gc11/oms11g/cfgtoollogs/omsca
Configuration assistant "OMS Configuration" Failed

In the latest log in /u01/app/oracle/product/gc11/oms11g/cfgtoollogs/cfgfw directory
I found the reason of the failure:

Check the OMS Configuration Assistant logs at:
/u01/app/oracle/product/gc11/oms11g/cfgtoollogs/omsca
INFO: oracle.sysman.top.oms:/u01/app/oracle/product/gc11/oms11g/bin/omsca:
line 40: /bin/java: No such file or directory INFO: oracle.sysman.top.oms:
Check the OMS Configuration Assistant logs at:
/u01/app/oracle/product/gc11/oms11g/cfgtoollogs/omsca

Further check of omsca code unveiled use of WebLogic configuration script
/u01/app/oracle/product/gc11/wlserver_10.3/common/bin/commEnv.sh
where there was code block responsible for JAVA_HOME and JAVA_VENDOR variables.

I simply put values for the variables to the script:
JAVA_HOME="/usr/java/jre1.6.0_20"
JAVA_VENDOR="Sun"

# Reset JAVA_HOME, JAVA_VENDOR and PRODUCTION_MODE unless JAVA_HOME
# and JAVA_VENDOR are pre-defined.
if [ -z "${JAVA_HOME}" -o -z "${JAVA_VENDOR}" ]; then
# Set up JAVA HOME
JAVA_HOME=""
# Set up JAVA VENDOR, possible values are
#Oracle, HP, IBM, Sun ...
JAVA_VENDOR=
# PRODUCTION_MODE, default to the development mode
PRODUCTION_MODE=""
fi

I reinstalled WebLogic, recreated database, removed entries from inventory.xml and started installation again.
It was finished successfully and I completed it with execution of allroot.sh script under root
from /u01/app/oracle/product/gc11/oms11g

Have a good day!

06:07 Fixing the Dreaded ‘Unable to locate an oracle.mk, proc.mk or other suitable *.mk’ Error in DBD::Oracle insalls (1703 Bytes) » The Pythian Blog

If you have ever had this message:

Unable to locate an oracle.mk, proc.mk or other suitable *.mk
file in your Oracle installation. (I looked in…)

It can be a very frustrating one to track down.

I have seen it across almost all platforms and versions of Oracle but what it boils down to is that you are quite simply:

1) Missing some part of your Oracle Client or
2) Part of your Oracle Client is some place where the Makefile.PL cannot find it or
3) Part of your Oracle Client is buried inside a resource or a symbolic link or
4) Oracle has moved or removed files that DBD::Oracle needs to compile.

Well what to do then??

There is no 100% answer for this problem but there seems to be two main scenarios where I have encountered this and have come up with somewhat of a solution:

1) If you are attempting to build against any of Oracle’s Instantclients you most likely have only have the basic install available. To correct this simply make sure you get the SDK and SQLPlus downloads as well and try compiling again.

2) If you are installing against an 11g instance (Oracle seems to have moved a number of things about depending on the platform) then you best bet would be to give this a try:

Perl Makefile.PL -l

which forces the compile to user the “direct link” method thus bypassing the *.mk files. Expect much more output and a longer compile time when you take this route but it should work exactly the same.

Cheeers
John Scoles

03:55 李雷和韩梅梅 (1512 Bytes) » Chanel [K]

虽然人物偏少,那两段叙事一般的歌词应该唠叨不同的人才对。

但是,它仍然触碰到了心底的那份柔软,多少年前在街边放肆奔跑的身影从背后嗖地掠过,大声欢笑着跑远,在这个夏日的傍晚,悄悄地让人望着夕阳的余晖发呆。

一转眼,我们都长大了。


01:27 Deal with Pgbouncer Error “closing because: Old V2 protocol not supported” (2272 Bytes) » DBA@SKY-MOBI
大概是在2008年的时候,公司的数据库已经初具规模了,为了更好的掌握数据库运行状态,做了一个服务器运行状态统计信息收集的系统。(当时对很多开源的工具还不太了解。现在想想发现和HQ,OACLE的grid control非常类似: 架构图如下: 分为三个层次: 第一个部分: agent,部署在被监控的服务器上面,收集实时的服务器运行状况,发送到repository. 我这里是使用ISQL来实现的(需要在被监控的服务器上安装unixODBC)。 第二部分: Repository,用于存储来自被监控的服务器发送过来的数据。另一个功能是提供给HTML服务器生成报告的历史数据。 我这里使用了一个PostgreSQL来作为Repository。 第三部分: Ctl层和HTML展现层,用于配置一些展现需要的信息,获取来自REPOSITORY的数据并在HTML上展现出来。 最近在一个测试环境中搭建了一套一样的环境。与之前不同的是,这次增加了一个PostgreSQL的连接池pgbouncer 1.3.3。 数据库使用的是PostgreSQL9beta1 。 一用就出现问题了,Agent去连接repository的时候报“[ISQL]ERROR: Could not SQLConnect”这样的错误。 到pgbouncer的日志中看到如下输出: 2010-06-04 15:35:04.733 15323 LOG C-0×998e490: (nodb)/(nouser)@xxx.xxx.xxx.xxx:17905 closing because: Old V2 protocol not supported (age=0) 2010-06-04 15:35:04.733 15323 WARNING C-0×998e490: (nodb)/(nouser)@xxx.xxx.xxx.xxx:17905 Pooler Error: Old V2 protocol not supported 从日志上分析,客户端发起请求使用的V2协议已经不被Pgbouncer支持了。 接下来马上绕过pgbouncer,agent直接和postgresql9连接,成功。 OK,找到问题的根源了,来解决它,毕竟PGBOUNCER还是要使用滴。 首先翻阅了PGBOUNCER的说明文档,最后在pgbouncer的维基发现有这么一句话”Supports protocol V3 only, so backend version must be >= 7.4.” 也就是说pgbouncer仅仅支持V3是没错的啦。 OK,那来看看ODBC的配置里面有没有可以指定协议版本的,又翻阅了一下unixODBC的手册,发现下面一段话: Protocol: Note that when using SSL connections this setting is ignored. 6.2: Forces driver to [...]

2010-06-03 Thu

12:34 Beginner/Expert (1 Bytes) » Oracle Scratchpad
07:09 A Universal SQL Client » DBA@SKY-MOBI

2010-06-02 Wed