All db and archive files reside on normal non-nfs file systems, but do to old space limitations we pointed the "database backup" command to write to an nfs mounted file system.
This was never an issue until we registered a new 10gR2 db in rman. All other db's backup fine, but the 10gR2 db results in the following error:
ORA-27054: NFS file system where the file is created or resides is not mounted with correct options
Additional information: 2
I've found some documentation for oracle on linux (we're Solaris also) that suggests special mount options are required:
specifically hard,rsize,wsize, and noac
But this doc is for linux, and appears to be geared to DB datafiles, and i'm not sure if they apply to Solaris and RMAN.
Our current mount options for the dir in question are:
Anybody know what we're missing specifically? And why its an issue in 10gR2, but not in previous releases?
Are the other databases backed up to the same nfs mount ? As far as I know oracle does support netapp, others I don't know about. The requirements for the backup pieces and the datafiles will be the same.
Our solution was to redirect RMAN to backup to a non-nfs mounted file system.
I mentioned a few article which detail recommended mount options for nfs file systems and Oracle. But they were for Linux systems, and not terribly specific.
Other than that we have not investigated the issue further, or seen any other informative explanations here, on metalink, or on OTN.
Sorry, I couldn't be of more help. When you resolve the issue, please post the solution here. I am very curious to know the specifics, but as of yet had not had the time to investigate further myself.
With these settings was the backup speed per sec the same or longer.
mount -o hard,rw,noac,rsize=32768,wsize=32768,suid,proto=tcp,vers=3
When I changed the mount options to rsize=32768,wsize=32768, hard, noac,
my backup speed when from 9MB per second to 316KB per second.
Plus all of my Oracle 9i backup to the same NFS mounts were slower too.
The problem is the same issue. Performance was very poor. A database backup to local disks took 3 minutes. The same backup to NFS mounted disk with rsize and wsize set to 32768, 16384, or 1048576 was 1 hour and 9 minutes.
Do you know a way to turn this checking off?
After many days and many test runs, Oracle Tech Support and I finally found the solution that work for me. rw,bg,intr,hard,timeo=600,wsize=32768,rsize=32768,nfsver=3,tcp
Now my backups to local disks and to NFS mounted disk are the same.
For the backup that took 3 minutes to local disks and the same backup to NFS mounted disk took 1 hour and 9 minutes, both now take 3 minutes.
This solution works with no adverse affect on my Oracle9i RMAN backups to the same NFS mounted disks.