ONTAP Discussions

Is anyone running Oracle VM on cDOT 8.2?

ronaldmajor
7,789 Views

We have a strong desire to run Oracle VM on cDOT 8.2 in a Flexpod environment which we already have in place.

All of the documentation I have found is for 7-mode filers.

I am wondering if there are any caveats or limitations to using NFS repositories or FCoE boot luns.

Also, should the OSTYPE for LUNs and igroups be Linux or XEN?

Any advise is greatly appreciated.

Thanks

1 ACCEPTED SOLUTION

nkarthik
7,789 Views

Hello RonaldMajor,

NetApp and Oracle has written the joint vendor OVM Best practice technical report(TR-3712), which will be released to puplic in couple of days in www.netapp.com/us/library/.

The OSTYPE for LUNs and igroups should be linux.

Regards,

Karthikeyan.N

View solution in original post

11 REPLIES 11

nkarthik
7,790 Views

Hello RonaldMajor,

NetApp and Oracle has written the joint vendor OVM Best practice technical report(TR-3712), which will be released to puplic in couple of days in www.netapp.com/us/library/.

The OSTYPE for LUNs and igroups should be linux.

Regards,

Karthikeyan.N

ronaldmajor
7,747 Views

Thank you for the quick reply.  I currently have TR-3712 which is for 7-mode.  I assume the new one will be an update for cDOT?

nkarthik
7,747 Views

Yes RonaldMajor, The TR specifically updated for cDOT.

 

Regards,

Karthikeyan.N

nkarthik@netapp.com

Cell: xxx-xxx-xxxx

 

edit: number removed

ronaldmajor
7,747 Views

We have been unable to get OVM Manager to work with the showmount plug-in for cDOT and thus register the NFS volumes.  I believe we are missing something and TR-3712 does not give specific instructions on how

to configure OVM Manager to use NFS volumes on cDOT.  I can install the plug-in and it runs correctly from the command line.  Since the plug-in communicates with a management port on the filer and the volume is accessed

from a data port, we are not sure how to configure the Generic NFS plug-in in OVM Manager.  Obviously it must work since the authors of TR-3712 did it.

Any ideas?

ronaldmajor
7,747 Views

I found the answer.

     The showmount plug-in should be installed on ALL OVM servers.

     The LIFs used to access (mount) the NFS repository volumes should have the mgmt firewall policy so the showmount plug-in can access with ONTAPI.

     The volume for the repository must be exported and mountable on ALL OVM servers.

     Make sure / is exported and is readonly! (Servers cannot mount anything if / is not exported.)

With these requirements met, normal procedures for filesystem management in OVM Manager can be followed.

nkarthik
7,747 Views

Great Ronald to find the solution and share with others.

Regards,

Karthikeyan.N

TULANIAN47
7,747 Views

Ronald,

Still having problems with the showmount plugin. We've done the four steps you list here but are still getting "Authorization Failed" from the plugin. I found this comment in NaServer.py:

    def set_admin_user(self, user, password):

        """Set the admin username and password.  At present 'user' must

    always be 'root'.

    """

Are you using root as the username?

Thanks,

Bill

ronaldmajor
7,747 Views

Bill,

You need to create an account on the StorageVM for the showmount tool to use. This is the SVM that contains the OVM repository.

Give the account vsadmin-readonly role and put the username and password for this account in the new showmount file as stated in the documentation.

I hope this helps.

TULANIAN47
7,747 Views

Ah, the vsadmin-readonly role was the missling link. Thanks!

Bill

TULANIAN47
5,815 Views

Another question here if I may. My storage admin has asked me to mount the volumes using specific LIFs. However, because the LIFs and volumes are members of the same vserver all of the volumes are visible behind all of the LIFs. OVM Manager doesn't seem to allow you to specify the IP for the mount if the storage is visible behind more than one and instead picks the last IP that was used to refresh the storage.

Page 35 of TR-3712 mentions adding a mount directly to the fstab on the servers but since the repositories aren't added to the fstab by OVM Manager I don't know how it will react (instead mounting the repos during the initialization of cluster services). Has anyone actually done this?

Thanks,

Bill

kjcreason
5,494 Views

>Page 35 of TR-3712 mentions adding a mount directly to the fstab on the servers but since the repositories aren't added to the fstab by OVM Manager I don't know how it will react (instead mounting the repos during the initialization of cluster services). Has anyone actually done this?

 

I am curious about this as well. I've been using iSCSI but would like to compare NFS for repositories, and also have the separate LIFS for management and data so the default file server access fails.

Does anyone have feedback on this? Anyone done any extending of the plugin or showmount fix to accomodate the separate lifs?



Public