A long thread name? Well the topic is vast, hence the long name. I’ve been dealing with Oracle EBS or applications R12 from last 9 years. Initially I was ONLY developing for the new infrastructure, that gradually changed to taking care of the whole instance.
Painfully, however definitely I did familiarize myself with Linux (RHEL) & the database, exposing myself to building systems with copies of EBS running for development & testing.
Our instance is approximately 650GB as on date, including both application stack and database & a cold backup is hardly 125GB in tar balls. I have attempted & succeeded to build the cloned instances on my home desktop machine many times, however the performance was a huge issue, forcing me to discard the setups quite often.
This time, I decided NOT to discard as my new desktop at home is a beast compared to my previous machine, that was neither less a best ;) & to figure out “something” that will address the “performance” bottleneck.
I created a new VM using Oracle VirtualBox with following specs
6 processors, 20GB memory & 2 fixed size VDI files (120GB, 600GB) respectively for application and database repositories. I was aware of a limitation already, I was setting up the VDI files on a Western Digital Green series 2TB drive! which spins at 5400RPM!
Well, everything went smooth & and I had the instance up and running in couple of hours time & this time, the response of the instance was awesome. I even boasted about finally winning over the “biatch” to my team & sat back feeling “too proud” for the moment.
Next day (Month End 30th April 2019)
I am all excited after figuring out a way to flush GNOME desktop environment & replace it with Xfce & new tricks…
Started the VM at home, started the EBS instance and tried to access the instance from the same machine. I couldn’t even get the login page…something was gravely wrong. I decided to check the performance monitor and found the following:
Slowly I was forced to recognize the terror! The Standard Concurrent manager was configured to process 25 requests at the same time with a cache of 5 & 30 seconds sleep between the requests. Our month end has a number of scheduled jobs + Gather schema statistics in the queue. My VM was breaking up with the I/O. My 2TB storage oriented disk was NOT spinning fast enough to provide the data for the processes & I was left with the BIGGEST question of the hour “Now what?”
I stopped the Concurrent manager, adjusted the processes to 5 for Standard manager & restarted the instance. Left the instance running whole night and 1st May morning, the instance was back to normal performance as whole the scheduled jobs were finished during the night.
Next day I added one 1TB SSD & moved the application & database VDI files over to it. I was able to get the login screen within 2-3 seconds once after the application started from the VM. I submitted number of create accounting and other resource hungry jobs, which were completed in few seconds time…
Now, my setup is ONLY for the sake of it. It doesn’t have many users, it is idle most of the time & almost every day I shutdown the desktop machine after a day’s usage. This might not be the scenario at a real TEST environment. You may need to implement archive logging & RMAN, those all requiring more space & faster access to storage. A Desktop has less resources & the ONLY positive element you are going to live with is the pleasure of building it & knowledge gained while fixing few new issues.
So, can you build a performance oriented R12 using Desktop environment, the short answer is yes. Does it worth the efforts? Well, definitely YOU are the ONE who have to answer it.
Follow the space & soon I will post a thread explaining the entire exercises. If you are in a hurry, you may refer this
The above article loosely explains how to clone R12 instance on Linux 7. However the same could be followed for Linux 6 as well (both RHEL & OEL)
I have 3 Windows 10 Professional licenses (Yes, paid licenses). 2 of the licenses are used in two hardware devices & the 3rd one I am using with a VirtualBox VM machine.
My semi server class desktop computer at home has approximately 4.5T storage, consolidated from different disks. My Windows10 VM is frequently factory reset to try out different software (Mostly Oracle software at uncertified environments) & the particular partition on which I had the VM was running out of space & I planned to move the VM & the disks to a different partition with bigger free space.
Scenario: Move “Windows 10” VM from L: drive to M: drive
My nightmare started when the copy process stopped abruptly with Windows popping an error window stating there were read errors while copying disk1.vdi & I attempted again to realize that the disk1.vdi is has bad sectors or clusters.
I hurried to confirm that the VM is still accessible by restarting the VM couple of times & to my greatest pleasure the VM did start, did shutdown properly and I was able to access the Oracle database installation.
A quick googling suggested that an attempt to repair the hard disks using “CHKDSK”, the old utility could fix the .vdi corruptions & without wasting more time I jumped in
At an elevated command prompt issued the command
> chkdsk L: /F
My partition is of 1TB & after 15-16 minutes the chkdsk completed with few messages like fixing some cluster information.
I tried to copy the disk1.vdi once again to M: drive, and this time the copy process managed almost 19GB out of 29.1GB and again presented me the read error.
Now, I was left with just one more option in front of me & it was to attempt recover the bad sectors, and there were chances that my virtual disk may get corrupt if the damages are more in numbers. knowing my VM is already Activated by Microsoft, I could try to reinstall Windows in a new virtual disk and retain the license as long as I am still using the other VM files as the id of the machine registered with Microsoft.
Please be informed, if you copy ONLY the vdi file, the activation will be nullified and you will be prompted to activate the copy of the Windows as soon as you start the new VM built using the copied VDI file!
So I proceeded to
>chkdsk L: /F /R
Windows suggested that it may take around 6 hrs for the repair process & I woke up next day morning to read multiple status messages about moving the corrupt sectors to new positions & some errors those could not be repaired.
I hurried to copy the disk once again & even though it seemed like forever, finally I was able to copy the disk1.vdi to M: drive!
So if you ever come across such a situation, give the above a try. If you are not cursed, most probably you will able to recover the virtual disk.
A year back, following the Oracle’s blog, got excited and engaged with a tedious task to setup a Vision instance for Oracle R12 12.1.3 using the templates.
Against all the warnings and negative responses, I did manage to setup Oracle VM Manager and other repositories using Oracle Virtual Box and put the Vision instance online.
The efforts were unjustified as the tasks involved were supposed to be carried out by a seasoned DBA with advance level Linux knowledge. Apparently I lost the appetite to continue with the solution and discarded the entire setup along with the data repositories.
which explained a different approach towards setting up a Vision instance using the same Oracle templates for R12 12.1.3 and decided to give it a try.
I must say, John Piwowar @ pythian has put everything in place, so that a novice like me could follow the instructions without making any kind of mistakes and get the vision instance up and running.
Once the vision instance was up and running, I started wondering whether I could try something further. Oracle’s template based approach was involving two independent virtual machines serving two different roles
One server providing the database node
Second server providing the application node
At work I am used to deal with a single server hosting both database and application nodes which makes few things like server administration, OS patching etc easier.
Hence I started pondering the possibilities of merging both database and application servers into a single server.
Right now, after around 48 hours, I have successfully managed to merge both the nodes into a single server (virtual) and online with the Vision instance & I am going to share the entire procedures in detail so that you can also!
PART-1 Collect and arrange your stuff
A desktop/laptop machine with minimum 4GB free RAM for the virtual machine. That means you must have 6-8GB base memory
Loads of free storage space. If your fixed disk doesn’t have minimum 500GB free disk space, I recommend you to buy a 1T external HDD (the read/write speed may be compromised, USB3 could be a minor exception)
Access to a very fast internet connection to download the templates from Oracle e-delivery portal
You have to download 11 files, 38GB by continuing
Pythian post describes a method to use wget to download all the files listed, using a file list input parameter, from a OSx/Linux host.
I downloaded the above files from a Windows machine, and must say it was a painful process.
PART-2 Convert the template files into .img and then .vdi files
I am not going to repeat the steps which are already clearly narrated with pythian article. Only suggest that, don’t use a Windows machine and 7zip to do all the unzip and joining the files. You use a Linux Virtual machine to do those jobs. It would be faster, and it would be the best. I created a Oracle Linux 6 VirtualBox to do those jobs by sharing the downloads folder with the Virtual Machine.
PART-3 Backup everything!
So once the img files are converted as .VDI files, the most import thing is to backup the .VDI files to somewhere else! Yes, you may corrupt the sources available inside the .VDI files while attempting to setup the Vision instance. If you don’t have a backup, you have to go through the painful process of unzipping, combining and converting the .img files to .VDI formats. Trust me, this exercise could easily run into 4-6 hours time, depending upon the efficiency of your hardware.
PART-4 1st Gear, slow start
Create a folder called “myvision” (name it anything)
Copy both System.vdi and ebs1211db.vdi (While converting the .img files I gave the .VDI files the same image names) from your database template folder or backup folder to “myvision” folder
The actual size of ebs1211db.vdi over Windows environment will be approximately 220GB, which will be few GB smaller over Linux environment.
As we are planning to host both the dB and Application tier from a single server, we MUST increase the size of ebs1211db.vdi
Let me explain why, prior going ahead
Oracle’s approach towards the VM setup was purely involving two servers(virtual) and both database & application files are placed on /u01 mount point
ie, Database server has a mount point called /u01 and so does the application server
All the clone scripts available for both database & applications servers have certain scripts hardcoded with /u01 as base path. Hence, one could get into useless activities of editing scripts to change the hardcoded base path values to adjust the cloning (Which was done successfully)
PART-5 2nd Gear, Resize ebs1211db.vdi
Fortunately, ebs1211db.vdi doesn’t have multiple partitions. Hence resizing the volume is pretty easier using Virtual Box command line tools & free linux tools like Gparted
Switch to command prompt (if you are on Windows 7 & 8 make it elevated)
Release ebs1211db.vdi from the Gparted Virtual Machine. Let us start creating a Virtual Machine using Oracle’s VirtualBox. Please note we are using the latest version of VirtualBox, if you are not, please update prior you move ahead with rest of the exercises
Please check the VM configurations I made for my Vision Instance Server with the below image
You can see that I have attached both System.vdi & ebs1211db.vdi files under SATA
Now, most important: You must have a .iso image or a physical DVD/CD for a recent release of Oracle/CentOS/RHEL linux, as we have to alter few attributes of the Virtual Machine we are going to put online.
Oracle has created the VM compiling a XEN kernel, which will not work under VirtualBox & you will fail to boot your Vision Instance VM, unless the kernel pointers are changed and recompiled.
Don’t worry it is pretty easy
Follow pythian article until the new kernel installation part
If you have just downloaded the templates from Oracle, you have the latest release of Oracle Linux embedded with System.img/.vdi file, and you don’t have to install a new kernel.
#rpm –qa | grep kernel
If your uek kernel version matches as one provided with the image or higher, you can safely move ahead with creating a new initrd entry. As root issue the following command
Reboot your Virtual Machine, after removing the Linux media from the CD rom drive (or detach the linux iso image)
If you were following pythian post attentively, your new virtual machine MUST boot, without throwing any errors and prompt you for a login
The default password for root is “ovsroot”
I hope you did change the /etc/fstab entry for “/u01” mounting point prior rebooting the server as mentioned last time and that you have changed the
/etc/sysconfig/oraclevm-template entry to RUN_TEMPLATE_CONF=YES
If you haven’t, please do & restart the Virtual Machine. Optionally you can set a static IP to your box, so that you can start using a ssh client like putty, which will genuinely make many things easier while setting up the server at later stages.
Once the server restarted, you will notice that Oracle will automatically start the cloning process based on the “RUN_TEMPLATE_CONF=YES” entry
Throughout, we were using a hostname “appvis”, hence all the instructions, workarounds will utilize the same hostname. Please adjust the hostname according your preferences
Enter static IP address: 192.168.0.30 Enter netmask: [255.255.255.0] Enter gateway: 192.168.0.1 Enter DNS server: 192.168.0.1
Shutting down interface eth0: [ OK ] Shutting down loopback interface: [ OK ]
Configuring network settings. IP configuration: Static IP address
Bringing up loopback interface: [ OK ] Bringing up interface eth0: [ OK ]
Enter hostname (e.g, host.domain.com): appvis.rhome.com
Network configuration changed successfully. IP configuration: Static IP address IP address: 192.168.0.30 Netmask: 255.255.255.0 Gateway: 192.168.0.1 DNS server: 192.168.0.1 Hostname: appvis.rhome.com
Database node related prompts
Enter the APPS password :apps
Provide the values required for creation of the new Database Context file.
Target System Hostname (virtual or normal) [appvis] :<<press ENTER>>
Target Instance is RAC (y/n) [n] : n<<press ENTER>>
Target System Database SID : VIS
Target System Base Directory : /u01/E-BIZ
Target System utl_file_dir Directory List : /usr/tmp
Number of DATA_TOP’s on the Target System  :<<press ENTER>>
Target System DATA_TOP Directory 1 [/u01/E-BIZ/db/apps_st/data] : /u01/E-BIZ/db/apps_st/data
Target System RDBMS ORACLE_HOME Directory [/u02/E-BIZ/db/tech_st/11.1.0] :/u01/E-BIZ/db/tech_st/22.214.171.124
Do you want to preserve the Display [atgtxk05:0.0] (y/n) : n
Target System Display [appvis:0.0] :<<press ENTER>>
Do you want the the target system to have the same port values as the source system (y/n) [y] ? : n
Target System Port Pool [0-99] : 42
Make a note of the System Port Pool you have entered here, as we must use the same while setting up the Application tier, in order to establish a connection to the database & perform the cloning process
Sit back and let the cloning process for database to complete. It may take a while. Once the cloning process is over, the database will be started.
After a successful completion of cloning, the server will come back to a logon prompt
login as “oracle” and the default password is “oracle”
You will be asked to change the password for user oracle and APPS, ignore the prompts this time.
PART-7 4thd Gear, Free on the roads, hacking the Application tier
As I mentioned, what we are going to do is a pure hack, against Oracle’s design structure.
The first step is to create a soft link, which is mandatory for cloning the Application tier
Login as root cd /usr/lib cp libXtst.so.6 libXtst.so.6.backup rm libXtst.so.6 ln -s /usr/X11R6/lib/libXtst.so.6.1 /usr/lib/libXtst.so.6
As prompted with the Database tier clone, a number of prompts you must answer, commonly the following
Enter the APPS password : apps Target System Hostname (virtual or normal) [appvis] : <ENTER to take default> Target System Database SID : VIS Target System Database Server Node [appvis] : gkar Target System Database Domain Name [rhome.com] : <ENTER to take default> Target System Base Directory : /u01/E-BIZ Target System Tools ORACLE_HOME Directory [/u01/E-BIZ/apps/tech_st/10.1.2] : <ENTER to take default> Target System Web ORACLE_HOME Directory [/u01/E-BIZ/apps/tech_st/10.1.3] : <ENTER to take default> Target System APPL_TOP Directory [/u01/E-BIZ/apps/apps_st/appl] : <ENTER to take default> Target System COMMON_TOP Directory [/u01/E-BIZ/apps/apps_st/comn] : <ENTER to take default> Target System Instance Home Directory [/u01/E-BIZ/inst] : <ENTER to take default> Target System Root Service [enabled] : <ENTER to take default> Target System Web Entry Point Services [enabled] : <ENTER to take default> Target System Web Application Services [enabled] : <ENTER to take default> Target System Batch Processing Services [enabled] : <ENTER to take default> Target System Other Services [disabled] : <ENTER to take default> Do you want to preserve the Display [atgtxk-09:0.0] (y/n) : n Target System Display [appvis:0.0] : <ENTER to take default> Do you want the the target system to have the same port values as the source system (y/n) [y] ? : n Target System Port Pool [0-99] : 42 UTL_FILE_DIR on database tier consists of the following directories. 1. /usr/tmp 2. /usr/tmp 3. /u01/E-BIZ/db/tech_st/126.96.36.199/appsutil/outbound/VIS_appvis 4. /usr/tmp Choose 1, default
The cloning will kick start now, and resume until 89% prior ending up with a FATAL error
–Minor update (if it is already NOT too late) 08-October-2013
I was trying to do a cloning & realized that the cloning process terminates with various errors and most of the errors were due to write permissions.
I was able to complete the cloning processes (dbTier & appsTier) individually by changing the permissions on folders like
and after the clone, resetting the ownership of /tmp back to root
chown –R root:root /tmp chmod –R 777 /tmp
Hence the below mentioned error “could be” addressed by changing few OS level rights as mentioned above.
This error is caused by missing “adautocfg.sh” file in the folder /u01/E-BIZ/inst/apps/VIS_appvis/admin/scripts which the auto configuration tries to access.
Without this file in this particular directory, the cloning process will never complete successfully. So we have to find a solution for it (I am not a APPS DBA and genuinely do not have any idea why the heck this file is missing in this particular location. All I could do is a guess work and it is towards avoiding the oracle template configuration wizard part, which initiates few functions. I’ll check it one of the following days and update this post accordingly)
and paste the following (Yes you can do it, just change the VIS_appvis with your CONTEXT name)
#!/bin/sh # dbdrv: none
# # $Header: adautocfg_ux.sh 120.4 2008/02/19 04:28:02 sbandla ship $ # # ############################################################### # # This file is automatically generated by AutoConfig. It will be read and # overwritten. If you were instructed to edit this file, or if you are not # able to use the settings created by AutoConfig, refer to Metalink Note # 387859.1 for assistance. # # ###############################################################
The cloning process must complete without throwing any further errors and you will be asked whether the application should be started by the end of the cloning process. You can go ahead with starting the application tier.
However, you will notice, the apache HTTP will fail to start, as the CONTEXT FILE will have entries, which were not properly put in place during our last clone process as it has ended up with a FATAL error.
Let us edit the CONTEXT file and correct those entries now.
(This situation arises only if the cloning is done as user root)
Incase if you are not sure about the http port, open up “VIS_appvis.xml” CONTEXT FILE and scan through, you will find it with url address area
Use sysadmin/sysadmin to logon and enjoy!!!
PART-8 5th Gear, Smooth Sailing, Restarting and online
As a system administrator I literally don’t like the idea of starting resource hungry services until the boxes are completed booted up and online. I prefer to start certain services myself and make sure that whatever were started manually completed the startup process successfully.
As root switch to “/etc/init.d” folder
chmod –x ebizdb
and you are done! The database will not startup next time while you are starting your Vision instance.
Go to these folders for starting database and application nodes
Day 3 – Adding Desktop to Oracle’s basic server (system.vdi)
I personally prefer to have a proper GUI, coming from an entirely Windows environment, that is a MUST for many times, regardless whether I am working with Linux or Windows systems
Here are the few things I have done to get the GNOME desktop installed with the basic Oracle linux Server embedded with the Oracle VM templates for Oracle R12 12.1.3
(Please make sure, from your Oracle linux box you have access to internet. Do you a ping www.google.com and confirm you are getting replies)
(Please shutdown both database and application prior proceeding)
(Please increase the size of System.vdi file PRIOR experimenting with any of the below exercises, and be careful while you would use Gparted to do it. Make a backup for system.vdi file and follow the thread provided above with the post to resize the partition (it is easy, I assure))