Download as pdf or txt
Download as pdf or txt
You are on page 1of 3

3/4/24, 16:28 Document 279393.

1
Copy right (c) 2024, Oracle. A ll rights reserv ed. Oracle Confidential.

Linux/NetApp: RHEL/SUSE Setup Recommendations for NetApp Filer Storage (Doc ID 279393.1)

In this Document

Purpose
Scope
Details
Single Instance
RAC Instances
10g RAC Notes
Description of All Related Mount Options
References

APPLIES TO:

Linux OS - Version 2.4.21 to 2.6.9-89


Linux x86
Linux x86-64 *** ***
Linux Kernel - Version: 2.4.21 to 2.6.9-89.EL

PURPOSE

This article will assist Linux system admins properly setup a NetApp Filer with Red Hat Linux AS/ES 2.1, RHEL 3.0, 4.0, and
SuSE Linux SLES8 and SLES9.

SCOPE

This article is for system administrators configuring NetApp file with Red Hat Linux AS/ES 2.1, RHEL 3.0, 4.0, and SuSE
Linux SLES8 and SLES9.

For more up-to-date recommendations on RHEL5/SLES10, please goto:


http://now.netapp.com/Knowledgebase/solutionarea.asp?id=kb7518

DETAILS

Single Instance

SLES9 , RHEL4 , RHEL3

rw,bg,hard,nointr,tcp,vers=3,timeo=600,rsize=32768,wsize=32768

RHAS2.1:

rw,bg,hard,nointr,tcp,vers=3,timeo=600,rsize=32768,wsize=32768

In /etc/modules.conf add:Oracle Release >= 10.1.0.5: Voting & OCR file requires Patch 4466428

options nfs nfs_uncached_io=1

https://support.oracle.com/epmos/faces/DocumentDisplay?_adf.ctrl-state=ayacsymf3_151&id=279393.1 1/3
3/4/24, 16:28 Document 279393.1
NOTE: Only set this for AS 2.1

RAC Instance s

SLES9, RHEL4, RHEL3 QU3 and later:

rw,bg,hard,nointr,tcp,vers=3,timeo=600,rsize=32768,wsize=32768,actimeo=0

RHAS2.1, RHEL3 pre QU3:

rw,bg,hard,nointr,tcp,vers=3,timeo=600,rsize=32768,wsize=32768,noac

And for AS 2.1 in /etc/modules.conf add:

options nfs nfs_uncached_io=1

NOTE: Only set this for AS 2.1

10g RAC Note s

Two options have been available for making sure that the OCR and Voting files are accessed with Direct I/O.

1. Oracle Relase <= 10.1.0.4: Voting & OCR file requiresPatch 4309443 and Patch 4466428, which opens these files
O_DIRECT, and the actimeo=0 mount option; patches are included in PSR 10.1.0.5 and 10.2.0.2.
2. In RHEL3, you could create a separate mount point for the voting file (quorum disk) and OCR, then use the following
mount options:

rw,bg,hard,nointr,rsize=32768,wsize=32768,tcp,vers=3,timeo=600,sync

But in RHEL4 this workaround is no longer valid with RHEL4 as the conservative caching algorithm was changed to a more
aggressive one

This is only for the mount from which the voting file/OCR is accessed. The voting file does not have to be in a separate
volume, it just has to be accessed by the hosts from a separate mount point using these options. Therefore it's best to use
option #1 above (2 patches + actimeo=0).

Use the following mount options for the rest of your datafiles, logs and control files:

rw,bg,hard,nointr,tcp,actimeo=0,vers=3,timeo=600,rsize=32768,wsize=32768

De scription of All Re late d M ount Options

option: rw
required for: all

option: fg OR bg
required for: all

option: noac
required for: RH AS21 after e.35
To enable O_DIRECT writes over NFS for kernel versions later than e.35, nfs must be loaded with the nfs_uncached_io=1
parameter
required for: RH EL3 pre U3 ensuring data cosistency with rh

NOTE: noac is the same as setting both the sync,actimeo=0 options in RHEL3

option: actimeo=0
required for: ALL RAC INSTALLS
Stems from Oracle behavior with autoextend, that the new filesize is not propagated to other nodes in the cluster. without

https://support.oracle.com/epmos/faces/DocumentDisplay?_adf.ctrl-state=ayacsymf3_151&id=279393.1 2/3
3/4/24, 16:28 Document 279393.1
this option, NFS will cache the old filesize and incorrect behavior will result.

NOTE: Using actimeo=0 disables all NFS attribute caching. This is required when you have the possiblity that the file will be
extended
and the other nodes not notified. Currently, Oracle depends on filesystem messaging to advertise the changed size of
datafiles, therefore this setting is necessary.

option: hard
required for: ALL ORACLE INSTALLS
Prevent NFS from returning "short write" errors. Short writes will cause the database to crash; otherwise will continue
retrying at timeo=NN intervals

option: nointr
required for: ALL ORACLE INSTALLS
Without this option, kill -9 will result in data corruption in oracle datafiles because the in-flight writes will be abruptly
terminated.

option: tcp
required for: ALL
Do not use UDP NFS for ANY REASON.

option: vers=3
required for: ALL
NFS v4 is a work in progress; there is no reason to run NFS v2.

option: rsize/wsize
required for: ALL
rsize and wsize should be set to equal to or a larger multiple of the Oracle block size to prevent fractured blocks.

option: nolock
required for: contraindicated for all
nolock is dangerous because it disables the only way for two separate db instances to recognize that they have mounted
the same datafiles. nolock provides no performance benefit/degradation, however from an administration standpoint, the
use of nolock can
result in datacorruption if dbf's get mounted simultaneously on multiple nodes.

REFERENCES
PATCH:4466428
PATCH:4309443
Didn't find what you are looking for?

https://support.oracle.com/epmos/faces/DocumentDisplay?_adf.ctrl-state=ayacsymf3_151&id=279393.1 3/3

You might also like