how to prepare new instance of existing AMI installed with oracle - Oracle on Amazon Web Services Cloud (Archived)

Hi All
I have created one centos instance from existing aws ami's and attached 2 ebs volumes each of 50gb, created raid 0 in between the both ebs volumes and attached the raid volume as /raidvol, installed entire oracle software on /raidvol partition, started the databse and everything is working fine.
i prepared the existing instance as AMI.
now i have taken snapshot of the above 2 ebs volumes and made them as volumes
now when i create an instance of the above ami i generally get the old instance settings wrt oracle, how do i change the settings to new instance so that i can attach the existing volumes and start oracle.
kindly suggest any procedure.
Anil 

What do you mean by 'instance settings'? Are you referring to initialization parameters, kernel parameters, or the database state?
Anything that you want to persist has to be on the EBS volumes. If you have /etc/oratab, /etc/oraInst.loc, oraInventory, etc. on the ephemeral storage, it will revert to AMI state on termination and restart. You should make all those things symbolic links to files in your EBS filesystem and rebundle the AMI.
Or you could use "boot from EBS".
Jeremiah Wilton
Blue Gecko, Inc.
[http://www.bluegecko.net|http://www.bluegecko.net] 

To store AMI, S3 storage is needed which is extra charge.
May be, you can do following
- Create AMI using pubic image
- Create new volume using EBS
- Move Oracle binaries (ORACLE_BASE) to EBS
- Create database with all files on EBS (e.g. admin files, db files, ..)
- Backup .profile on EBS (which has ORACLE_HOME, ORACLE_SID etc.)
- Terminate and recreate AMI
- Attach your EBS volume
- Delete pre-configured binaries and restore your .profile
HTH,
nilesh 

nilesh_patel wrote:
To store AMI, S3 storage is needed which is extra charge.For a 1GB AMI, it is $0.15 / month.
- Create AMI using pubic image
- Create new volume using EBS
- Move Oracle binaries (ORACLE_BASE) to EBS
- Create database with all files on EBS (e.g. admin files, db files, ..)
- Backup .profile on EBS (which has ORACLE_HOME, ORACLE_SID etc.)
- Terminate and recreate AMI
- Attach your EBS volume
- Delete pre-configured binaries and restore your .profileThe pieces that you need to persist are the /etc/oratab, the (s)pfile and the passwordfile. You don't need to copy the whole ORACLE_HOME. You just have to fix up symbolic links on those two places in the OEL community AMI to point to your persistent copies on EBS.
Jeremiah Wilton
Blue Gecko, Inc.
[http://www.bluegecko.net|http://www.bluegecko.net]

Related

Moving to new servers/host DB/Portal/eBus

We are moving our hosting to new servers. I have been looking at different ways to try and acomplish this. I have read through some posts and documents trying to determine the easiest way to do this. We are hosted on RedHat 4.8 and moving to the same. There are no mount points (/u0x) set up on the new servers. It is a fresh install of RedHat. Since we are moving to a new servers and a new host server names, etc will be different.
DB server (10.2.04)
From what I see it would be easiest to install the DB software and use RMAN backup to pull in the tables? This requires you ensure the patch levels match, is there an easier way to just move the code base without installing and working the patch level up to the same?
Infra & Portal 10.1.4.1
Can the code tree just be moved over or will it need to be installed and updated?
Apps 11.5.10.2
Since there are no mount points set up to match what is currently available, can they be created and move the code base over? I have read some on Rapid Clone but it seems to indicate there is a code base already installed.
I have read through
Oracle® Database Backup and Recovery Advanced User's Guide
http://blogs.oracle.com/AlejandroVargas/gems/RMANDUPDBPRACTICE.pdf
and the 11i Rapid Clone document
But I want to make sure we are on the right track, there are multiple ways it looks like it CAN happen, but I would prefer the easiest.
Thanks to all 
Hi,
DB server (10.2.04)
From what I see it would be easiest to install the DB software and use RMAN backup to pull in the tables? This requires you ensure the patch levels match, is there an easier way to just move the code base without installing and working the patch level up to the same?You could simply use Rapid Clone to clone the database tier node -- See (Note: 230672.1 - Cloning Oracle Applications Release 11i with Rapid Clone) and (Note: 799735.1 - Rapid Clone Documentation Resources, Release 11i and 12) for details. No previous database installation is required on the target database node, just run preclone on the source node, copy the files, and run postclone on the target node.
Infra & Portal 10.1.4.1
Can the code tree just be moved over or will it need to be installed and updated?Cloning the application server is covered in "Oracle® Application Server Administrator's Guide".
Cloning Application Server Middle-Tier Instances
http://download.oracle.com/docs/cd/B14099_19/core.1012/b13995/cloning.htm
Apps 11.5.10.2
Since there are no mount points set up to match what is currently available, can they be created and move the code base over? I have read some on Rapid Clone but it seems to indicate there is a code base already installed.Just follow the same steps explained under the database cloning section above for the application tier node. Again, you do not have to install any 11i software on the target nodes before doing this clone. However, you will just need to install all the OS pre-req. software and packages on the target nodes before running postclone.
Regards,
Hussein 
Thanks for the input, it sounds a little less daunting. What about the mount point issue, will post clone create them or do I need to have them defined the same prior to running the post clone process? 
Hi,
What about the mount point issue, will post clone create them or do I need to have them defined the same prior to running the post clone process?The mount points should be defined in the system before copying the files and running postclone. Please note that the mount points names do not have to macth the ones you have on the source nodes.
Regards,
Hussein 
And for my final stupid question.....for now :)
What is the logic behind the /u0x points with defined sizes? Currently we just have a large / drive space and that is it. Every oracle installation I have seen has seperate mount points, is there a document that explains why and prefered settings for these?
Thanks a million as usual. 
Hi,
There is no preferred settings, and it is up to you to specify the number/name of those mount points -- For example, you could create a mount point called /u01 and have all the files (application/database) under it, or create two mount points /u01 where you can place the application files, and /u02 where you can copy the databasse files.
Regards,
Hussein 
Since we are moving to blade servers connected to a SAN, having the control files seperated to different logical volumes for disk IO pusposes wouldn't even be applicable in this case would it?
As far as the DB server I think I am going to create a /u01 for oracle home and /u02 for the dbf files along with another area to use for archive logs/auditing files and staging when needed since it is no longer a matter of Optimal Flexible Architecture IO performance issues.
Thank you for your input on everything, once again you saved me hours of trying to find it myself. :) 
Hi Toolman;
Thank you for your input on everything, once again you saved me hours of trying to find it myself. :)I agree wiht you, we are soo lucky to have Hussein Sawwan in this forum site. As i said many times He is BEST OF THE BEST
Regard
Helios 
Hi,
Thank you guys for the kind words.
Regards,
Hussein

Recommended File System Paritioning for EBS R12!

Hi All,
I'm in the process of cloning a Master Instance into Production Instance. Currently, in Master instance(both APPS & DB) we have a single file system parition on which, entire database/application is stored. In a nomal practice, it is always better have the files(for instance, DataFiles/Control Files) spread across different partitions(for better management).
Just wondering, what would be the suggested(per Industry standards) file system partitioning methodology for Oracle EBS R12 production Instance.
Thanks in Advance!
Mahesh 
Hi;
I'm in the process of cloning a Master Instance into Production Instance. Currently, in Master instance(both APPS & DB) we have a single file system parition on which, entire database/application is stored. In a nomal practice, it is always better have the files(for instance, DataFiles/Control Files) spread across different partitions(for better management).
Just wondering, what would be the suggested(per Industry standards) file system partitioning methodology for Oracle EBS R12 production Instance. It depends which RAID system do you use. You can create mount point like
/u01 << for INST_TOp
/u02 << appl_st
/u03 << tech_st
but if those path use same disk on your RAID even this method wont be helpful..
Regard
Helios 
I'm in the process of cloning a Master Instance into Production Instance. Currently, in Master instance(both APPS & DB) we have a single file system parition on which, entire database/application is stored. In a nomal practice, it is always better have the files(for instance, DataFiles/Control Files) spread across different partitions(for better management).
if your cloning, best way is have same size of mounts in test servers.
but best practice is keep side db file system and apps binaries in diffrent mount points.
Just wondering, what would be the suggested(per Industry standards) file system partitioning methodology for Oracle EBS R12 production Instance.
best is check this oracle document below.
http://download.oracle.com/docs/cd/B34956_01/current/acrobat/120oaig.pdf
page no : 25 Disk Space Requirements

Migration of Storage R12

Hi All, EBS R12.1.3 with PCP 2 nodes shared appl_topDB R11.2.0.2 RAC 2 nodes ASMOS RHEL 5.5 Currently we are on EVA4400. And now new we migrating to 3PAR storage.  At this stage , as an OracleAppsDBA what will be the responsibilties during this migration.I mean what role dba has in this migration. Thanks Mapps
In theory, and assuming you want to change as least as possible to the Oracle application, there's not a lot to do, other than stopping the complete application (including database), seeing the disks being changed to their new location (DF would probably tell you), confirm all data and structures are still there, and start up. If you want to use this occasion to change any path, or you must change any path (for whatever reason), there's more work to be done.
I would suggest talking to the sysadmin(s) and storage admin(s), just to make sure any necessary mount options are set, and to coordinate testing, especially if the NFS gateway is changing.

+ prefix in datafiles names on Linux for Oracle 11 g

Hi, We are working on a 11 g database on Linux (RHEL system) and i noticed a curious thing in the names of datafiles (in v$datafile) the names of all the datafiles begin with +/DATA/..i don't know this syntax or notation, what does it mean ? Thanks a lot
jef_2802 wrote:
 
Hi,
 
We are working on a 11 g database on Linux (RHEL system) and i noticed a curious thing in the names of datafiles (in v$datafile)
the names of all the datafiles begin with +/DATA/..i don't know this syntax or notation, what does it mean ?
 
Thanks a lot
https://www.google.com/webhp?hl=en&tab=ww#hl=en&q=oracle+asm
hi, thanks for the replyI did not know this technology for Oracle.I will read
The "+" sign represents the starting character from where the navigation would start. Think about it the same like "C:\" which we have in Windows. That's the character from where the navigation starts in Windows, right. In ASM, the same thing is done using the + . HTHAman....
Starting with Oracle 10g, Oracle introduced the +ASM instance. ASM is a volume manager and file system that was designed for Oracle database files. It is the Oracle recommended standard that replaces the use of OS file systems, like Windows NTFS or Linux EXT, for example. Keep in mind however, that from a database perspective, there is no I/O difference whether database files are accessed with or without ASM. The ASM instance does not cache, translate or manage actual data content. The +ASM instance transfers file extent metadata to the database SGA for direct access to the actual data. There are several advantages, for example, you do not need to worry about database and filesystem block size. The database simply defines the db block size.
It points your storage system is in ASM (asm diskgroup) not an OS file system location .If you want to see this location just connect to asm instance and use asmcmd > ls 

Moving PROD - DB 11gR1 + EBS 12.1.3 from Virtual to physical

Hi,DB - 11.1.0.7  entEBS - 12.1.3 DB size - 550gb OS - RHEL 5.4 64bitKERNEL - 2.6.18-155.el5 x86_64 Physical server - HP ProLiant DL380 G6Storage - 6 disk Raid 5Downtime - Can take 24-48hrs (weekend) also several times during weekend before actual implementation after testing.Directory and Partition mount point - will be created same as source (check below point 3) Note -  1) DB+EBS on same server i.e Single server2) V2P will not be done by Virtualization team. They say that doesn't happen. So consider it Has no option.3) Parition mount points - It will be created same to avoid rman restore and many other things like customization, ftp and stuff don't want to change data files and4) Many customization are there So DBUA option and stuff will create more problems than help.5) Same architecture same Os version same kernel.. Just Virtual to physical so drivers and stuff OS related will create issues. NO UPGRADE WILL BE DONE DURING MIGRATION Also No Transformation.6) Same SID,HOSTNAME will be preferred to avoid updating several clients knowledge. (Optional) Task : Want to move from Virtual to Physical (stated above). I would like to know the easiest method to migrate to new server. I need advice for the same. OS1) OS issue - Have to install new OS but 11.1.0.7 is not Compatible with 5.4 + linux or 2.6.18+ kernel. So stuck with 5.4Also the current Physical server doesn't consist of any partition or os. I am thinking of using Oracle Linux 5.4 in place of RHEL 5.4 because of few issues on like ISO availability, repo, etc2) Want to know the easiest method to copy user,groups,ftp, etc in short of the setting IF THERE IS ANY METHOD than please state.3) Want you guys use to backup the OS ? like clonezilla is there any other alternative for Physical server (not virtual)4) Crucial : Storage is there any better way to create parition ? i mean alignment, block size etc ?  I have read that this particular kernel and db version doesn't support much of the OS disk enhancement parameter or settings. Can anyone tell me how to create an efficient Partition with parted or fdisk. DB+EBS1) What is the simplest method to migrate to new server via COLD BACKUP? What are the exact step to do COLD BACKUP restore ?2) If suppose (due to datafiles moving) time and network speed factor becomes bottleneck than RMAN tape backup (of data) but take Coldbackup copy of Oracle home and EBS will suffix ? if yes than how exactly ?3) Will i need to relink the Oraclehome binary ?  Will i have to run Autoconfig on both after relink ?4) Have to do anything similar things for EBS ?5) In this coldbackup scenario can i also do Incremental RMAN backup application to DB after selecting ACTUAL config and doing full restore but before starting apply incremental.6) Will standby solution will be better in this situation ?  Benchmarking1) What will be the easiest method to benchmark DISK io completely as per Oracle Database   What i did before postingHave read different method of restore but i am not able to find this particular case as of mien (using cold backup).Have read about RMAN different methods (Liked it)Standby methods, (Liked it)CTAS (not applicable)Import export (not applicable)Golden gate (Additional License not applicable)Read about Disk Alignment via official and non official sourcesCold backup Cloning Why i made this threadEvery thing i did is theory (never did v2p) i want some one who did it practically to comment and help in plan out So to avoid error and time lag.I also needs tips and tricks from Gurus in this situation. I hope i didn't miss out anything. Message was edited by: 3310241 Note - point 5 and 6
I would do this with Rapid Clone. That has to be a more reliable technique than trying to invent your own method.
thx for the reply But i have other questions to ! like the pratition, the linux etc
You haven't bothered to research Rapid Clone at all, have you?
I have used rapid clone many times for Cloning the prod env to test and uat but in this case the OS is not there it is totally blank server. So i asked .Also I think it is not said as cloning but migration that is where i am confused as to it should be exact copy of the PROD env . I am just saying my side not blaming you as i misunderstood that term.We do clone with rapidclone only preclone post clone stuff but i thought there might be something different in migration that is why i created.Also i mentioned in the first post that i studied cloning didn't mentioned rapid clone word though. I would be very glad if you can comment on the OS stuff as you have made clear that rapid clone will be perfect for this situation but regarding the OS stuff i am still uncertain.I hope i didn't offend you. Thx for the generous reply.      

Categories

Resources