"John Szakacs" wrote:
>
> How do you backup your cluster?
>
> We are goign to implement a simple two node NFS cluster in a couple weeks.
>I cannot find any useful info on backing up the cluster resources using
Netbackup.
>Our setup will have one NFS resource on each of the two nodes. Since we
>must specify a CLIENT in the bp.conf, how do we handle a failover?
>If the NFS resource fails from one machine to another then wont that
>failed over resource get a FULL backup since it has not been backed up
>on that particular client? Not too bad but we will have 1/2TByte
>resources.
>
>Thanks in advance for any help.
>Cheers,
>john

John,

I have the same problem.
My workaround is based on the fact that the virtual IP and the storage (VM
disk group)
are allways on the same node.
I configure a new netbackup client that point to the service group IP. In
this case
the backup will be with this client regardless in which node it is running.
The problem is that the CLIENT_NAME entry in bp.conf on the client is only
one entry.
If the client name is the virtual IP then will be a problem to backup the
local disks.
I make a test and the CLIENT_NAME in bp.conf has no effect weather i backup
with
the maint IP or virtual IP.
The only restriction is that you cannot backup simmultaneoslly both nodes
( virtual node and maint IP node because you have only one binary client
inst and the client
cannot act as two diff clients in the same time.

If you know about a better solution to backup a VCS cluster with Netbackup
--PLEASE
LET ME KNOW

Pablo Horenstein wrote:

> "John Szakacs" wrote:
> >
> > How do you backup your cluster?
> >
> > We are goign to implement a simple two node NFS cluster in a couple weeks.
> >I cannot find any useful info on backing up the cluster resources using
> Netbackup.
> >Our setup will have one NFS resource on each of the two nodes. Since we
> >must specify a CLIENT in the bp.conf, how do we handle a failover?
> >If the NFS resource fails from one machine to another then wont that
> >failed over resource get a FULL backup since it has not been backed up
> >on that particular client? Not too bad but we will have 1/2TByte
> >resources.
> >
> >Thanks in advance for any help.
> >Cheers,
> >john
>
> John,
>
> I have the same problem.
> My workaround is based on the fact that the virtual IP and the storage (VM
> disk group)
> are allways on the same node.
> I configure a new netbackup client that point to the service group IP. In
> this case
> the backup will be with this client regardless in which node it is running.
> The problem is that the CLIENT_NAME entry in bp.conf on the client is only
> one entry.
> If the client name is the virtual IP then will be a problem to backup the
> local disks.
> I make a test and the CLIENT_NAME in bp.conf has no effect weather i backup
> with
> the maint IP or virtual IP.
> The only restriction is that you cannot backup simmultaneoslly both nodes
> ( virtual node and maint IP node because you have only one binary client
> inst and the client
> cannot act as two diff clients in the same time.
>
> If you know about a better solution to backup a VCS cluster with Netbackup
> --PLEASE
> LET ME KNOW

HI
Don't bother to think about cluster config in your NetBackup configuration

Just backu the nodes with a class with only the Cluster filesystems configured.
If you want to use NetBackup to take down a database ore likewise you can use
the scripting facility
in NetBackup to check if the node own the cluster group
example:
SYSTEM=`hastat -summary | grep Oracle| grep `uname -n`
if ( $SYSTEM != 0 );then
hagrp -freeze Oracle -sys `uname -n`
fi

If this isn't aplicable then the NetBackup will backup only the mount points if
the group is
currently on the other machine

You can do a restore from node1 to node 2 if you want to


Regards Patrik

Hi,

You will need to install netbackup on your shared disk using a vitrual IP
and virtual name in the event that hte server fails over and requires the
indexes.

You must create a link for the /usr/openv/netbackup default with your new
location inorder for the media to be available unpon failover.



Patrik Linder wrote:
>
>Pablo Horenstein wrote:
>
>> "John Szakacs" wrote:
>> >
>> > How do you backup your cluster?
>> >
>> > We are goign to implement a simple two node NFS cluster in a couple
weeks.
>> >I cannot find any useful info on backing up the cluster resources using
>> Netbackup.
>> >Our setup will have one NFS resource on each of the two nodes. Since
we
>> >must specify a CLIENT in the bp.conf, how do we handle a failover?
>> >If the NFS resource fails from one machine to another then wont that
>> >failed over resource get a FULL backup since it has not been backed up
>> >on that particular client? Not too bad but we will have 1/2TByte
>> >resources.
>> >
>> >Thanks in advance for any help.
>> >Cheers,
>> >john
>>
>> John,
>>
>> I have the same problem.
>> My workaround is based on the fact that the virtual IP and the storage
(VM
>> disk group)
>> are allways on the same node.
>> I configure a new netbackup client that point to the service group IP.
In
>> this case
>> the backup will be with this client regardless in which node it is running.
>> The problem is that the CLIENT_NAME entry in bp.conf on the client is
only
>> one entry.
>> If the client name is the virtual IP then will be a problem to backup
the
>> local disks.
>> I make a test and the CLIENT_NAME in bp.conf has no effect weather i backup
>> with
>> the maint IP or virtual IP.
>> The only restriction is that you cannot backup simmultaneoslly both nodes
>> ( virtual node and maint IP node because you have only one binary client
>> inst and the client
>> cannot act as two diff clients in the same time.
>>
>> If you know about a better solution to backup a VCS cluster with Netbackup
>> --PLEASE
>> LET ME KNOW
>
>HI
>Don't bother to think about cluster config in your NetBackup configuration
>
>Just backu the nodes with a class with only the Cluster filesystems configured.
>If you want to use NetBackup to take down a database ore likewise you can
use
>the scripting facility
>in NetBackup to check if the node own the cluster group
>example:
>SYSTEM=`hastat -summary | grep Oracle| grep `uname -n`
>if ( $SYSTEM != 0 );then
>hagrp -freeze Oracle -sys `uname -n`
>fi
>
>If this isn't aplicable then the NetBackup will backup only the mount points
if
>the group is
>currently on the other machine
>
>You can do a restore from node1 to node 2 if you want to
>
>
>Regards Patrik
>
>
>
>Content-Description: Card for Patrik Linder
>
>begin:vcard
>n:Linder;Patrik
>tel;fax:+46 31 750 2001
>tel;work:+46 31 750 2067
>x-mozilla-html:FALSE
>org:

Dimension
>AB
>adr:;;Mölndalsvägen 85 ;Gothenburg;;412 85;Sweden
>version:2.1
>email;internetatrik.linder@gbg.dimension.se
>title:Consultant
>x-mozilla-cpt:;31720
>fn:Patrik Linder
>end:vcard
>
>