Comments for Oracle 11g Release 1 RAC On Linux Using NFS
Monowar said...Thanks for this article but I am not sure why it is mentioned version 10gr2 to the following location:
Install the Clusterware Software
Unzip the clusterware and database software.
Tim... said...Sorry. It's a bad cut&paste. I've corrected it now.
Arif Gulzar said...RAC1:
IP Address eth0: 192.168.2.102 (public address)
IP Address eth1: 192.168.0.102 (private address)
TIM It should be RAC2 not RAC1 right?
Tim... said...Yes. I've corrected it now. Looks like a need an editor. :)
Krishna said...Hi Tim,
I am bit confused here.
I was under the impression that DNFS is an NFS client embedded into oracle. So why would we need to mount the filesystems through the OS for Oracle to see it?
Should oracle not be shooting rpc directly to the Filer without any OS intervention at all?
One of the ways that Direct NFS identifies what it can access is to see what is currently NFS mounted. If Oracle needs to write to these NFS mounts is accessed them directly through Direct NFS. It's just using the OS mounts as a source on information.
Rememeber also, you need the mount points available before the Direct NFS software has actually been loaded, so you can install the clusterware.
mkdir -p /u01/app/crs/product/11.1.0/crs
mkdir -p /u01/app/oracle/product/11.1.0/db_1
mkdir -p mkdir -p /u01/oradata
mkdir -p mkdir -p /u01/shared_config
The first two lines are ok, but it looks like the last two lines both have a typo.
this is a great install guide, Thanks!
Chintan said...Dear Tim,
Really a Nice Work Done for all DBA's We can start installing and working with Oracle 11g RAC
Thank You Very Much.
Sean said...I ran into a problem that others have had on 11g. The GV$CLUSTER_INTERCONNECTS view shows that one of the private interconnects is using the wrong ethernet interface.
Vishal Desai said...When I start creating Database using DBCA it hangs at 2% and I cannot cd into my nfs file systems.
Tim... said...Questions in the forum, not in the comments!
Sounds like you've not setup your NFS mounts properly. If you can't cd into them, there's no point starting the installation.
mash said...thanks Tim,
database creation causes system to hang at 1%
still checking nfs mount options
check out (for nfs options)
I can verify that noac was required for my setup
below is my /etc/fstab
nas1:/shared_config /home/oracle/shared_config nfs rw,bg,hard,nointr,tcp,noac,nfsvers=3,timeo=600,rsize=32768,wsize=32768,actimeo=0 0 0
nas1:/shared_crs /home/oracle/crs nfs rw,bg,hard,nointr,tcp,noac,nfsvers=3,timeo=600,rsize=32768,wsize=32768,actimeo=0 0 0
nas1:/shared_home /home/oracle/db nfs rw,bg,hard,nointr,tcp,noac,nfsvers=3,timeo=600,rsize=32768,wsize=32768,actimeo=0 0 0
nas1:/shared_data /home/oracle/oradata nfs rw,bg,hard,nointr,tcp,noac,nfsvers=3,timeo=600,rsize=32768,wsize=32768,actimeo=0 0 0
Val said...Can I use Solaris-86 instead of Linux?
Igor said...Thanks for very useful guide.
As far as I understand for the correct "direct nfs client" usage we need to export shared_data with insecure attribute:
so we can see data in v$dnfs_files, v$dnfs_channels views and i/o improvements.
One question: why Oracle write speed remains very slow (~ 3 mB/sec) despite of OS write speed is quite high (~ 30 mB/sec) on the same nfs volume (/shared_data)?
Rob said...Hi Tim,
Thanks again for another great article. I was wondering, is there any reason why you left the hangcheck-timer part out (it was there in the 10g article).
Set the hangcheck kernel module parameters by adding the following line to the /etc/modprobe.conf file.
options hangcheck-timer hangcheck_tick=30 hangcheck_margin=180To load the module immediately, execute "modprobe -v hangcheck-timer".
wak said...iyo lah pe ko kater...
JohnH said...Does each computer need to have two network cards so that a public IP address and a private IP address can be set? Do the two computers connect to each other via two routers: one for public IP, and another for private IPs? Thank you.
Ravi said...Hi Tim,
I assume that I have to edit /etc/exports on both servers and then edit fstab on both servers. I tried but the NFS mounts are coming up fine on RAC1 but on RAC2, the NFS mount just finishes but the df doesn't show me any NFS mounted filesystem on RAC2. NFS deamon is running and there are no typos in the fstab or exports on both servers.
If I do a ps -ef I still see the mount /u01/* processes running and they never finish.
Raju said...Under Direct NFS Client section ..
mv libodm11.so libodm11.so_stub
ln -s libnfsodm11.so libodm11.so
cp libodm11.so libodm11.so_stub
ln -s libnfsodm11.so libodm11.so
No. It is meant to be a move. I want to move the original file (libodm11.so) and replace it by a symbolic link to libnfsodm11.so.
Steve said...Bear in mind, the enterprise edition install, by itself, installs a whole bundle of options which require additional licenses.
And this /isn't/ covered by the OTN license.
Barbara said...This is amazing Tim... Thank you very much for sharing this on the web...
DO NOT ask technical questions here! They will be deleted!
These comments should relate to the contents of a specific article. Constructive criticism is good. Advertising and offensive comments are bad and will be deleted!