[Ocfs-users] problem with OCFS label

[Wu, Hai]

Raj,
Thanks. My worry is about the entries in /etc/fstab file. The configuration would be the same across all nodes in the RAC. Say:
/dev/emcpowerd   /oradata/data3   _netdev, ...

If emcpowerd was changed to emcpowerc, then this mounting would fail. We have to manually modify that file then. I want to use LABEL to work around that problem, and I know that won't work now. It seems it has to be done manually. Please let me know if I got it wrong.

Your suggestion of copying powerpath config file across nodes is a good one. To do 'powermt save', then copy the file. I am still not sure where that file is at this point.

Thanks,
Hai

-----Original Message-----
From:	Raj [mailto:sanstorage at gmail.com]
Sent:	Thu 9/1/2005 7:12 AM
To:	Wu, Hai
Cc:	Joel Becker; ocfs-users at oss.oracle.com
Subject:	Re: [Ocfs-users] problem with OCFS label
You would not need to worry about that even if you add/remove LUNs. Powerpath maintains a consistent "Logical device ID" across nodes. Even if your nodes scans the same LUN as say for example empowerc on node1 and emcpowerd on node2, thier Logical device ID will be the same. Hope that helps.

Raj


On 8/31/05, Wu, Hai < hxwu at tribune.com> wrote:

Raj,
I am using PowerPath 4.3.2/RHEL ES3U4/OCFSv1/CX700. For the same node, are you saying I don't need to worry about adding/removing LUNs? I thought the device number would change, say from emcpowerd to emcpowerc, if we were to remove emcpowera.

Thanks,
Hai

-----Original Message-----
From:   Raj [mailto: sanstorage at gmail.com]
Sent:   Wed 8/31/2005 10:48 PM
To:     Wu, Hai
Cc:     Joel Becker; ocfs-users at oss.oracle.com
Subject:        Re: [Ocfs-users] problem with OCFS label
Wu,

OCFS label does not work with EMC powerdevices. EMC powerdevices uses a logical ID number which knows the path to the correct LUNs on the backend. Compare those on each nodes and they will be the same. Do a powermt display dev=all and you find that long number which will be the same across nodes. If you need a consistent look and feel across all the nodes you will need to copy the Powerpath config file across nodes and then reboot. What version of PowerPath are you using?

Hope that helps.

Raj


On 8/31/05, Wu, Hai < hxwu at tribune.com> wrote:

Thanks Joel. If I can't use OCFS label, then how can I make sure I am still mounting the same emcpower device if we need to add/remove powerpath LUNs? I know that devlabel would not work in this case. (And IBM this its own way of dealing with this problem, not sure about EMC in this case).

Thanks,
Hai

-----Original Message-----
From:   Joel Becker [mailto: Joel.Becker at oracle.com]
Sent:   Wed 8/31/2005 8:28 PM
To:     Wu, Hai
Cc:     ocfs-users at oss.oracle.com
Subject:        Re: [Ocfs-users] problem with OCFS label
On Wed, Aug 31, 2005 at 06:42:20PM -0500, Wu, Hai wrote:
> But when I tried to mount this ocfs filesystem by label, it gave me the following error. 
> # mount -L data13 /oradata/data13
> mount: the label data13 occurs on both /dev/emcpowerp1 and /dev/sdad1 - not mounted

        This isn't a bug, this is a common problem with multipath.
mount(8) does not know that /dev/emcpowerp1 and /dev/sdad1 are the same 
drive.  So it cannot decide which to use.  You must mount by device name
in this case.

Joel

--

"Gone to plant a weeping willow
On the bank's green edge it will roll, roll, roll.
Sing a lulaby beside the waters. 
Lovers come and go, the river roll, roll, rolls."

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


_______________________________________________
Ocfs-users mailing list
Ocfs-users at oss.oracle.com
<  <http://oss.oracle.com/mailman/listinfo/ocfs-users> http://oss.oracle.com/mailman/listinfo/ocfs-users> http://oss.oracle.com/mailman/listinfo/ocfs-users