[prev in list] [next in list] [prev in thread] [next in thread] 

List:       ocfs2-users
Subject:    Re: [Ocfs2-users] Ocfs2-users Digest, Vol 69, Issue 21
From:       David Johle <djohle () industrialinfo ! com>
Date:       2009-09-28 20:24:13
Message-ID: 200909282024.n8SKOces018484 () mix ! industrialinfo ! com
[Download RAW message or body]

At 01:31 PM 9/23/2009, ocfs2-users-request@oracle.com wrote:
>Date: Wed, 23 Sep 2009 12:27:15 +0200
>From: Marc Grimme <grimme@atix.de>
>Subject: Re: [Ocfs2-users] CDSL .vs. Bind mounts?
>To: ocfs2-users@oss.oracle.com
>Message-ID: <200909231227.15539.grimme@atix.de>
>Content-Type: text/plain;  charset="iso-8859-1"
>
>As we are making shareing the "/"-Filesystem among multiple nodes we 
>are using
>CDSL as part of the concept.
>What we do is we create
><mount-point>/cluster/<nodeid>
><mount-point>/cdsl.local -> /cluster/nodeid (via bind mount)
>
>Then say you would like to make /opt/application/file1 hostdependent do the
>following
>mkdir -p /cluster/cdsl/[1-n]/$(dirname /opt/application/file)
>cp -a
><mount-point>/opt/application/file /cluster/cdsl/[1-n]/$(dirname 
>/opt/application/file)
>ln -s <mount-point>/opt/application/file
><mount-point>/cdsl.local/opt/application/file
>
>That's it and it works pretty well.
>Hope this helps.
>This concept keeps all file that are hostdep and reshared
>in /cluster/cdsl/.. .
>Regards
>Marc.


So I only need one bind mount on each node no matter how many files 
and/or directories are host specific.
And can use this approach for single files as well as whole directories.
And any node can still access another node's copy if necessary since 
they are all still on the shared volume.

I like that approach, it's very clean and still retains all the 
"benefits" of CDSLs.  Nice!

_______________________________________________
Ocfs2-users mailing list
Ocfs2-users@oss.oracle.com
http://oss.oracle.com/mailman/listinfo/ocfs2-users
[prev in list] [next in list] [prev in thread] [next in thread]