OPB (Oracle PowerBytes)

It seems there is a bug that can cause an ORA-00600: internal error code, arguments: [4881], [], [], [], [], [], [], []

Seems to happen once in a while when you have your database configured with ftp and/or http and your user(s) try to login but fail on the credentials.

Now, there was already a patch 12329328 available for HP-UX Itanium, 10.2.0.5 (extended support!) but for Linux x86-64, 11.2.0.3.0 and 11.2.0.3.4 they were released recently. The bug is supposed to be fixed in the upcoming 11.2.0.4 too.

Also see Note: ORA-600 [4881] When Using FTP Client To Connect To The XDB Database Repository [ID 1530585.1]

We’ve run into the situation that we had to find a way to upgrade our 12.1.0.1.0 agents after a successful upgrade of OMS, from 12.1.0.1.0 to 12.1.0.2.0. Why? We have set up our own way of agent deployment.

Therefor, it was necessary to set the password for our oracle user on all nodes. I retrieved the relevant nodes that contain the agents by issuing:

/path/to/oracle/mid02/oms/bin/emcli get_targets|grep 3872|grep P

where “P” stands for our production environment.

After retrieving the list I used the node information to 1) create and run a script to set the password, after 2) agent upgrades from OEM12c (Setup > Manage Cloud Control > Upgrade Agents) run 3) root.sh where applicable and last but not least 4) remove the password again, assuming the agent runs under oracle:

ssh -q -o “StrictHostKeyChecking no” dba_user@host.domain.nl “sudo echo $PASSWORD | sudo passwd oracle –stdin”

Assuming that dba_user can passwordless ssh to all the relevant nodes.

Same goes for the root.sh script:

ssh -q -o “StrictHostKeyChecking no” dba_user@host.domain.nl “sudo /path/to/oracle/agent12c/core/12.1.0.2.0/root.sh”

and for the password removal, obviously:

ssh -q -o “StrictHostKeyChecking no” dba_user@host.domain.nl “sudo passwd -d oracle”

The ORA-01276 is caused by a bug (4309649), ref. note ID 1454534.1.
Oracle tries to use the REUSE option and, even without the *.control_files= option set (in a spfile configuration), it hits the bug, resulting in ORA-01276.

Oracle suggests as stated workaround: “If duplicate RMAN command fails with spfile, then use init.ora for the duplicate database instance.” The bug is fixed in 11g.

After testing myself I learned that, when using OMF, duplication indeed only is succesfull without the control_files reference in the pfile.

Hi,

So, what’s happening?

I am running a RMAN duplicate.
Auxiliary instance has spfile with
*.control_files=’/path/to/<SID>/controlfile/o1_mf_6nw2wfke_.ctl’,'/backup_path/to/<SID>/controlfile/o1_mf_9ns7wsq2_.ctl’#Oracle managed file
Oracle runs into ORA-01276: Cannot add file /backup_path/to/<SID>/controlfile/o1_mf_9ns7wsq2_.ctl. File has an Oracle Managed Files file name.

Workaround for now is to manually create pfile, delete control_file entries and create spfile. Then run a create controlfile script and recover manually.

This happens right after RMAN finishes restoring:
<SID>: restore complete, elapsed time: 06:51:41 Finished restore at 14-MAY-12 sql statement: CREATE CONTROLFILE REUSE SET DATABASE “<SID>” RESETLOGS ARCHIVELOG
released channel: C1
RMAN-00571: ===========================================================
RMAN-00569: =============== ERROR MESSAGE STACK FOLLOWS ===============
RMAN-00571: ===========================================================
RMAN-03002: failure of Duplicate Db command at 05/14/2012 05:48:38
RMAN-06136: ORACLE error from auxiliary database: ORA-01503: CREATE CONTROLFILE failed
ORA-01276: Cannot add file /backup_path/to/<SID>/controlfile/o1_mf_9ns7wsq2_.ctl. File has an Oracle Managed Files file name.
Now, I would assume that, given the fact that Oracle places omf names in the spfile with the remark #Oracle managed file, it should be able to recognize that and work around it.

Do I really have to manually alter an pfile to delete the *.control_files= option, recreate spfile, startup nomount and start my duplicate?

Please comment on this if you have suggestions, thanks!

So, playing around with an 11 database on oel6. Sweet! But, whatever I tried, I couldn’t get the /dev/shm mounted after boot. Thanks to a search on the Internet I ran into a topic in OTN Discussion Forums.

It states that there probably is a bug (https://www.redhat.com/archives/rhelv6-list/2011-February/msg00081.html).

Adding it appropriately to you fstab and adding the command /bin/mount -o remount /dev/shm in /etc/rc.d/rc.local will solve your problem.

What:
http://collaborate12.com/ (IOUG, OAUG, Quest)
When:
April 22-26, 2012
Where:
Mandalay Bay Convention Center
Las Vegas, Nevada, USA

#876 – Security Boot Camp: Secure Your Database in a Single Day

Date:
Tuesday, April 24
10:45 am – 11:45 am

Location:
Palm D

Product Line:
Applicable to All

Tracks:
Database, Manageability, Security, Risk and Compliance

Description:
What does it take to make the database secure, to comply with regulations and mandates like SOX and PCI? You may be amazed to know that by taking some simple steps you can perform in a single day, you can attain 60% of the requirements and make your database a secure place, as you will see in this session with demos, examples and code samples to take home.

Arup: thanks for sharing

Presentation
Scripts

Hi,

A long time ago I ran into the problem that I didn’t receive any push notifications anymore when my iPhone was connected to the internet through 3G. Once it was connected through wifi there was no problem.

So, I didn’t get any Facebook, iMessages, Whatsapp, etc.

Googling from time to time to find a solution. Finally, I found an Apple forum where someone gave the solution:

Go to Settings –> General –> Profiles

As you see, the Profile is called Test. In my case is was a profile with a certificate from the previous provider (I switch providers, a while ago).

This was the problem! All I had to do was delete the profile with the certificate from the previous provider and from that moment I was able to receive push notification on 3G again! Yay!!!

Hi,

I recently created a script wrapped around RMAN which does the following:

• delete all datafiles from a duplicate (auxiliary) directory (Fast Recovery Area)
• startup the duplicate (auxiliary) instance in nomount phase
• start a duplicate from target database to duplicate (auxiliary) database

Two things that had to be taken care of are at first:

1. add the (UR=A)(in tnsnames.ora on target side) to allow connections to blocked services (see My Oracle Support ID 362656.1)

2. issue an alter system register; to register the database service with the listener when starting the duplicate after a shutdown abort; rm -rf /dir/SID/datafile/* ; startup nomount

3. use password file for authentication

Both databases use Fast Recovery Area (FRA) and Oracle Managed Files (OMF). All this is automated in a script, wrapped around RMAN, and every time the script ran into an error:

ORA-19906: recovery target incarnation changed during recovery

So, what is happening here?

This is typically caused by older archivelogs stored in the Fast Recovery Area (FRA) on the auxiliary system.

During the recovery phase, rman will catalog all files found in the FRA and if there are older archivelogs (eg. from a previous duplicate attempt and different incarnation), the incarnation is changed in the controlfile, hence the error.

So my problem was solved when also doing a rm -rf /dir/SID/FRA/archivelogs/* (deleting the achivelogs, before starting the duplicate)

Download Oracle Linux (without the requirement of a support subscription) from http://edelivery.oracle.com/linux

• Log in as user root
• Download the yum repo file fromhttp://public-yum.oracle.com/

# cd /etc/yum.repos.d
# wget http://public-yum.oracle.com/public-yum-ol6.repo

If you want, you can edit the repo file and enable other repositories, e.g. enable [ol6_UEK_latest] by just setting enabled=1 in the file with a text editor.

Run yum repolist to show the registered channels and you see we are including everything including the latest published RPMs.

Now you can just run yum update and any time Oracle releases new security errata or bugfix errata for OL6, they will be posted and you will automatically get them. It’s very easy, very convenient and actually very cool. Oracle does a lot more than just build OL RPMs and distribute them, they have a very comprehensive test farm where they test the packages extensively.

Tip!!!

Make it easy to install Oracle 11g on Oracle Linux 6 : yum install oracle-rdbms-server-11gR2-preinstall (also on public-yum) and done. Note: it doesn’t install the database nor the required software.

This is what the 11gR2 preinstall does (x86_64 only):

• Causes the download and installation of various software packages and specific versions needed for database installation, with package dependencies resolved via yum
• Creates the user oracle and the groups oinstall and dba, which are the defaults used during database installation
• Modifies kernel parameters in /etc/sysctl.conf to change settings for shared memory, semaphores, the maximum number of file descriptors, and so on
• Sets hard and soft shell resource limits in /etc/security/limits.conf, such as the number of open files, the number of processes, and stack size to the minimum required based on the Oracle Database 11g Release 2 Server installation requirements
• Sets numa=off in the kernel boot parameters for x86_64 machines

Source: Lenz Grimmer (https://blogs.oracle.com/linux/entry/oracle_rdbms_server_11gr2_pre)

Thanks go out to: Wim Coekaerts (He is is the Senior Vice President of Linux and Virtualization Engineering for Oracle. He is responsible for Oracle’s complete desktop to data center virtualization product line and the Oracle Linux support program. )