NetBackup Enterprise 5.1 and RHEL 3.0 Clustered Client Restor - Veritas Net Backup

This is a discussion on NetBackup Enterprise 5.1 and RHEL 3.0 Clustered Client Restor - Veritas Net Backup ; I am using NetBackup Enterprise 5.1 MP2 on the master server. I have 2 Red Hat Enterprise Linux 3.0 clients that are active/passive clustered. I installed the Netbackup 5.1 client on each of the nodes in the cluster, selecting Linux ...

+ Reply to Thread
Results 1 to 4 of 4

Thread: NetBackup Enterprise 5.1 and RHEL 3.0 Clustered Client Restor

  1. NetBackup Enterprise 5.1 and RHEL 3.0 Clustered Client Restor


    I am using NetBackup Enterprise 5.1 MP2 on the master server. I have 2 Red
    Hat Enterprise Linux 3.0 clients that are active/passive clustered. I installed
    the Netbackup 5.1 client on each of the nodes in the cluster, selecting Linux
    2.4 from the drop down list of clients. I backup using the virtual server
    cluster name (not one of cluster server names) and this works wonderfully.
    BUT when i simulate a failover and the passive node now becomes the active
    node and the shared date disk is now on this node, the restore puts the data
    on the original active node. It doesn't even use the virtual cluster name.
    If I completely shutdown that first node, the restore keeps trying to put
    it back there instead of going through the virtual cluster name. It should
    make a difference which node in the cluster has the shared disk. The restore
    should put it back to where the active node AND the shared data disk are,
    correct??? It's not doing that. Any ideas???

  2. Re: NetBackup Enterprise 5.1 and RHEL 3.0 Clustered Client Restor


    Diane,
    The recommendation when backing up clustered servers is to have two different
    policies so that you will be able to recover the data.

    Have one policy using the client node names to backup the O/S data.

    Have another policy using the vitural cluster name to backup the shared data.

    That way when you fail over and need to restore the virtual data on the shared
    disks, it will be restored regardless of which system is the active node.


    You should review the NetBackup High Availability Guide for more suggestions/details,
    etc.

    Deb

    "Diane" wrote:
    >
    >I am using NetBackup Enterprise 5.1 MP2 on the master server. I have 2 Red
    >Hat Enterprise Linux 3.0 clients that are active/passive clustered. I installed
    >the Netbackup 5.1 client on each of the nodes in the cluster, selecting

    Linux
    >2.4 from the drop down list of clients. I backup using the virtual server
    >cluster name (not one of cluster server names) and this works wonderfully.
    >BUT when i simulate a failover and the passive node now becomes the active
    >node and the shared date disk is now on this node, the restore puts the

    data
    >on the original active node. It doesn't even use the virtual cluster name.
    >If I completely shutdown that first node, the restore keeps trying to put
    >it back there instead of going through the virtual cluster name. It should
    >make a difference which node in the cluster has the shared disk. The restore
    >should put it back to where the active node AND the shared data disk are,
    >correct??? It's not doing that. Any ideas???



  3. Re: NetBackup Enterprise 5.1 and RHEL 3.0 Clustered Client Restor



    Diane wrote:

    >
    > I am using NetBackup Enterprise 5.1 MP2 on the master server. I have 2 Red
    > Hat Enterprise Linux 3.0 clients that are active/passive clustered. I
    > installed the Netbackup 5.1 client on each of the nodes in the cluster,
    > selecting Linux 2.4 from the drop down list of clients. I backup using the
    > virtual server cluster name (not one of cluster server names) and this
    > works wonderfully. BUT when i simulate a failover and the passive node now
    > becomes the active node and the shared date disk is now on this node, the
    > restore puts the data on the original active node. It doesn't even use the
    > virtual cluster name. If I completely shutdown that first node, the
    > restore keeps trying to put it back there instead of going through the
    > virtual cluster name. It should make a difference which node in the
    > cluster has the shared disk. The restore should put it back to where the
    > active node AND the shared data disk are, correct??? It's not doing that.
    > Any ideas???


    Hello,

    I have the same problem: 2 RedHat-Cluster nodes backed up by NetBackup6.
    Each cluster service has its own shared data and a service ip address.

    I setup according your recommendation a separate policy and client with the
    name/ip of the service. The first backup run ok, then I failed over the
    service to the other node and run another backup. But then I have
    encountered the following warning/problem:

    >Warning bpbrm (pid=28992) from client news: WRN - Old TIR

    file /usr/openv/netbackup/tir_info/mnt/news/NetBackup_file_info.SharedData
    is missing. Backing up everything in /mnt/news

    (/mnt/news is my shared data)
    Should'nt the tir_info file also be shared?
    I didn't find documentation on how to setup this correctly.
    Thanks for any hint.

    Hans-Peter


  4. Re: NetBackup Enterprise 5.1 and RHEL 3.0 Clustered Client Restor


    "Deb Wilmot" wrote:
    >
    >Diane,
    > The recommendation when backing up clustered servers is to have two different
    >policies so that you will be able to recover the data.
    >
    >Have one policy using the client node names to backup the O/S data.
    >
    >Have another policy using the vitural cluster name to backup the shared

    data.
    >
    >That way when you fail over and need to restore the virtual data on the

    shared
    >disks, it will be restored regardless of which system is the active node.
    >
    >
    >You should review the NetBackup High Availability Guide for more suggestions/details,
    >etc.
    >
    >Deb


    Deb,
    You don't understand what I'm doing. First, I do not need to back up the
    OS, so I do not need 2 additional policies created. I only need to backup
    the shared disk. Now, with that in mind, I can backup and restore SUCCESSFULLY
    (using the virtual cluster name) when the shared disk is on Server 1, the
    active server in the cluster. When I failover the shared disk to the second
    server, which we'll call Server 2, which was previously the passive server
    but is now the active server, I can backup the shared disk SUCCESSFULLY,
    once again, using the virtual cluster name. The problem comes when I now
    try to restore back to Server 2, using the virtual cluster name. It keeps
    trying to restore USING THE ACTUAL SERVER NAME instead of the virtual cluster
    name, back to Server 1. Where does it get that info? Why does it try to restore
    back to the server itself? No where does that name exist from when it was
    backed up. Do you see what I'm saying? And by the way, the information in
    that document you are referring to, is for a Master Server in a Unix cluster,
    not for a clustered client that Veritas is trying to backup. Believe me,
    I already looked through all the documentation.

+ Reply to Thread