[Ocfs2-users] could not start cluster stack

[Joel Becker]

On Thu, Sep 15, 2005 at 10:38:11AM +0300, Yuval Baruch wrote:
> The problem was caused since OCFS2 installation created a directory under 
> /lib/modules,
> which didn't match the kernel version I have ( 2.6.9-11 ) the directory 
> OCFS2 created was 2.6.9-11.0.0.10.3.EL.
> To bypass it I copied the modules to my directory 2.6.9-11.EL I think it is 
> a bug.

	No, it is an error on your part.  You must match the kernel
version with the driver.  So, if you have ocfs2-2.6.9-11.0.0.10.3.EL,
you need kernel 2.6.9-11.0.0.10.3.EL and vice versa.

> Another problem I am dealing with is that OCFS2 modules can't be loaded.
> When I ran dmesg I saw the following message:"ocfs2: unkown symbol 
> generic_drop_inode".

	This is completely expected.  2.6.9-11.EL is missing this
symbol.  That is precisely why we don't ship an ocfs2 driver for
2.6.9-11.EL.
	Go to http://oss.oracle.com/~mfasheh/ocfs2_rpms/.  There you will
find packages for the 2.6.9-11.0.0.10.3.EL kernel.  If you have an
Oracle support contract, this kernel is supported by Oracle and thus by
Red Hat.

Joel

-- 

Life's Little Instruction Book #173

	"Be kinder than necessary."

Joel Becker
Senior Member of Technical Staff
Oracle
E-mail: joel.becker at oracle.com
Phone: (650) 506-8127