<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
<meta content="text/html;charset=UTF-8" http-equiv="Content-Type">
<title></title>
</head>
<body bgcolor="#ffffcc" text="#000066">
<tt>Martin,<br>
<br>
See my answers below:<br>
</tt><br>
Martin Norland wrote:
<blockquote
cite="midB1A6D74D09CBC444B7246C4CCA92D6DD09DDEB@EXCHANGE1.ms.rhoworld.com"
type="cite">
<pre wrap="">I'm trying to install 10.2.0.1 (and then patch up to 10.2.0.3). The
first step, of course , is to install CRS - but I'm getting the
following error when trying to specify the OCR location:
"The location /u02/ocr, entered for the Oracle Cluster Registry (OCR) is
a directory. Specify a shared raw partition or cluster file system file
that is visible by the same name on all the nodes in the cluster."
</pre>
</blockquote>
The message here is clear. You are specifying a directory and not a
file. Best thing to do would be to create a subdirectory under the
mount point (/u02) and then the OCR file. I may be wrong, but I think
there is a but that prevents the CRS to create file directly on the
mount point.<br>
<blockquote
cite="midB1A6D74D09CBC444B7246C4CCA92D6DD09DDEB@EXCHANGE1.ms.rhoworld.com"
type="cite">
<pre wrap="">The intention is to have 3 small partitions for the {2} OCR and {3} CSS
mirror(s), but I can't seem to get Oracle to acknowledge they're
clustered. The clustered filesystems work fine, and cluvfy recognizes
them as shared.
=========
=== relevant lines from /etc/fstab:
/dev/sdc1 /u02 ocfs2
_netdev,datavolume,nointr 0 0
/dev/sdc2 /u03 ocfs2
_netdev,datavolume,nointr 0 0
/dev/sdc3 /u04 ocfs2
_netdev,datavolume,nointr 0 0
</pre>
</blockquote>
Make sure the device are mounted on the same mount point on both nodes.
Better use LABEL=<volume label> instead of the device. That will
prevent problems in case of device name changing.<br>
Also, make sure you can create a file on one node and see the same file
on the other node. The CRS install script will do the same and compare
them to make sure the devices are shared.<br>
<blockquote
cite="midB1A6D74D09CBC444B7246C4CCA92D6DD09DDEB@EXCHANGE1.ms.rhoworld.com"
type="cite">
<pre wrap="">=== Excerpt from `./runcluvfy.sh stage -post hwos -n rac01,rac02
-verbose`
Checking shared storage accessibility...
WARNING:
Unable to determine the sharedness of /dev/sda on nodes:
rac02,rac01
Disk Sharing Nodes (2 in count)
------------------------------------ ------------------------
/dev/sdb rac02 rac01
/dev/sdc rac02 rac01
=== Permissions
# ls -ald /u02 /u02/ocr/ /u02/css/
drwxrwx--- 6 oracle dba 4096 Mar 6 17:47 /u02
drwxrwx--- 2 oracle dba 4096 Mar 5 20:09 /u02/css/
drwxrwx--- 2 root dba 4096 Mar 5 20:08 /u02/ocr/
=========
It's worth noting that I was mounting these through powerpath, but
backed off to /dev/sd identifiers in an attempt to get Oracle to
recognize them (since cluvfy wasn't acknowledging /dev/emcpower's). I'd
originally intended to mount these off /u0#/cd#/, but also backed down
in attempt to get Oracle to spot them.
Any thoughts or any glaring holes in my setup that this noob is missing?
I didn't find anything on metalink related to this error, other than
recommendations to use raw devices (recommendations for a different
configuration).
Thanks,
-M
</pre>
</blockquote>
<br>
<pre class="moz-signature" cols="72">--
Regards,
Marcos Eduardo Matsunaga
Oracle USA
Linux Engineering
7453 TG Lee Blvd | Email : <a class="moz-txt-link-abbreviated" href="mailto:Marcos.Matsunaga@oracle.com">Marcos.Matsunaga@oracle.com</a>
Orlando, FL 32822 | Phone/Fax : (407)458-1710
A crisis is when you can't say "Let's forget the whole thing."
---------------------------------------------------------------------------------------
The statements and opinions expressed here are my own
and do not necessarily represent those of Oracle Corporation.
</pre>
</body>
</html>