I want to install a multi-node 184.108.40.206 system on 2 servers. The O.S. will be RED HAT 4 update 4. Now i have SAN storage and i want to create into it a shared area for the apps system to implement the SHARED APPLICATION TIER option.
What is the best file system FORMAT to choose?
Can i use OCFS2 ?
Can i share the application file system on the OCFS2 format without any problems with READ/WRITE operations on same files ?
What is the best practise ?
Please note that i'm not talking about RAC here.. i only have 2 servers ( SERVER WITH DB, ADMIN AND CONCURRENT and the other will be for FORMS AND WEB )
I just want to get benefit of the SHARED APPLIATION TIER FILE SYSTEM option.
Actually my friend i checked for that on Metalink and i found that OCFS2 is not certified with Oracle Ebusiness. ASM is certified with Ebusiness but only with 10G R1 OR R12 databases.
Anyway i think i will go with OCFS2 and mount the applications teir files to the shared storage on SAN (OCFS2 format).
If anybody tried that before please let me know...i'm talking about the applications files not DB files.
This question really has nothing to do with Oracle Clusterware so you have posted in the wrong forum. The best place to look for a solution would be the MAA
Also OCFS on linux questions are probably best in the linux forums as it is a linux opensource software.
Did you try OCFS for EBS application server and not DB? Did it work, are you supported?
The guy who suggested the MAA site for this type of discussion, there is nothing on that site on the specific question being asked here. Don't get too focused on where the question is posted, it is the content and importance of the question that is of importance.
Finally which approach you took? NFS filesytem could be an option but i think if the NFS server goes down then all nodes will be impacted, please correct me if i am wrong. Would appreciate if you guys throw some more light on this issue.I am on Oracle 12.0.2.
With a 2 node non-RAC configuration, I am assuming 1 node is the database node and the other is the Application tier, or at least WEB+FORMS. In this configuration, if either node goes down, you don’t have an app anymore, so the single point of failure for NFS is not an issue.
In our current configuration, we have 1 DB node and 2 App nodes. We NFS mount from the DB node to the App nodes, because if we DB node goes down, again we have no application. But if just one App node goes down, the other App node can continue to work.
but if you had a RAC db then you could survive a single db or single app failure but not with NFS on the db server. hence it would be better to go for a cluster solution if oracle would get around to certifying any of them for this purpose. as of now nothing except for NFS is certified which is a major downer seeing as this is a perfect file for a cluster filesystem.
Setting the record straight -
Oracle documents including - 11i Clusterware install and ocfs2 best practices - clearly state you can use ocfs2 for shared binaries - with this in mind - an R12 install with ocfs2 as shared file system for application tier will crap out with APache - there is a work around - I am going to try a GFS - since I do not want NFS - and see if I get the same error ----
FOR THOSE WHO JUST DON'T GET IT - WE ARE TALKING APPLICATION BINARIES - NOT THE DATABASE