RMAN backups of HP-UX dedicated storage node fail with error "Dedicated storage node leda2 cannot save remote data from client cora127n1.melkweg.tld"

Hello Martijn,

As you know Dedicated Storage Node (DSN) is used to store local data ( the client  the storage node is attaches to) . So ensure firstly you have seperate licenses for  DSN.

Secondly ensure that all the dns and hosts files ( name resolution ) is working well.

Also make sure that aliases are set properly, May be there is a mismatch where same box is seen as different one may be because it is multihomed or hostname not translated to FQDN.

Hi Bahaa,

Thanks for replying. We have 12 dedicated storage node licenses and 11 dedicated storage nodes. So license-wise I think we're okay. We don't use the hosts file and forward and reverse lookup of the backup server works fine on the HP-UX nodes. The HP-UX storage nodes are in a different DNS zone as the backup server. However we never had any problems before with this and like I said the other HP-UX cluster is in the same DNS zone and there the backup is fine. We might have some sort of breakthrough. I changed 2 things yesterday. I changed the remote access rights on the failing storage nodes. I entered the other storage node to the remote access rights of the first storage node.

storage node1

*@storagenode2

*@storagenode2.local

storage node2

*@storagenode1

*@storagenode1.local

And also I changed storage node1 from a dedicated to a shared storage node. All the backups of storage node2 succeeded and some of the backups of storage node1 failed with error:

ORA-19506: failed to create sequential file,, parms=""

ORA-27028: skgfqcre: sbtbackup returned error

ORA-19511: Error received from media manager layer, error text:

   Direct saves are not enabled. (0:5:2)

So I changed storage node1 back to a dedicated storage node. I requested to start some RMAN backups of storage node1 and they are running now.

Hello Martijn,

Did you try to take filesystem backups on the dedicated storage node ? and what was the results ?

Thanks,

Ahmed Bahaa

Hi Ahmed,

The file system backup works fine. We have no problems with the file system backup.

Regards,

Martijn

Hi,

If you want to use Dedicated Storage Nodes for physical hosts (lan free backup for cluster host) then you have to configure cluster host (virtual)  for Oracle backup as client in Networker and put curphyhost under client Storage node resource.

Then you will have correct licensing, now you are using 2 NMO licenses for one for each physical host, and you need only one if you configure cluster host

The licensing part is true only if you have active-passive cluster.

BR

Hi BR,

Here's the client config for both physical nodes of the HP-UX cluster. There's no client config of a virtual (cluster) resource. We only backup the file systems of the physical nodes and the databases are backed up with oracle RMAN.

C:\Windows\system32>nsradmin
NetWorker administration program.
Use the "help" command for help.
nsradmin> show
Will show all attributes
nsradmin> print type: nsr client; name: leda1
                        type: NSR client;
                        name: leda1;
                      server: cora127;
                   client id: \
d60d4028-00000004-48db8a93-48db8e9b-00090000-0a748a7f;
            scheduled backup: Enabled;
                     comment: "CDS:FS:HPUX";
             Save operations: ;
            archive services: Disabled;
                    schedule: Full Every Monday;
               browse policy: 3weeks;
            retention policy: 3weeks;
                  statistics: elapsed = 347581, index size (KB) = 296295,
                              amount used (KB) = 296295, entries = 1847917;
                   directive: Unix standard directives;
                       group: ASD-hpux;
                    save set: /, /var, /usr, /stand, /opt, /home, /d10,
                              /d70/orabck;
  Backup renamed directories: Disabled;
                    priority: 500;
   File inactivity threshold: 0;
File inactivity alert threshold: 0;
               remote access: *@pollux, *@pollux.sfb, *@oberon1,
                              *@oberon1.sfb, *@oberon2, *@oberon2.sfb,
                              *@leda2, *@leda2.sfb;
                 remote user: ;
                    password: ;
              backup command: ;
     application information: ;
                        ndmp: No;
             NDMP array name: ;
       De-duplication backup: No;
         De-duplication node: ;
         Probe resource name: ;
              virtual client: No;
               physical host: ;
           Proxy backup type: ;
           Proxy backup host: ;
             executable path: ;
    server network interface: ;
                     aliases: leda1, leda1.sfb;
                  index path: ;
          owner notification: ;
                 parallelism: 4;
               archive users: ;
               storage nodes: leda1;
       recover storage nodes: ;
         clone storage nodes: ;
                  hard links: Disabled;
             short filenames: Disabled;
                         BMR: Disabled;
                 BMR options: ;
                 backup type: ;
              client OS type: HpOS;
                        CPUs: 2;
           NetWorker version: LNMs_2007.Build.135;
              enabler in use: Yes;
       licensed applications: "NetWorker Module for Oracle, Unix Client/1";
               licensed PSPs: ;

nsradmin> print type: nsr client; name: leda2
                        type: NSR client;
                        name: leda2;
                      server: cora127;
                   client id: \
9bcb14c5-00000004-490977fc-490977fb-05010000-0a748a7f;
            scheduled backup: Enabled;
                     comment: "CDS:FS:HPUX";
             Save operations: ;
            archive services: Disabled;
                    schedule: Full Every Monday;
               browse policy: 3weeks;
            retention policy: 3weeks;
                  statistics: elapsed = 347620, index size (KB) = 223252,
                              amount used (KB) = 223252, entries = 1379407;
                   directive: Unix standard directives;
                       group: ASD-hpux;
                    save set: /, /var, /usr, /stand, /opt, /home, /d10,
                              /d70/orabck;
  Backup renamed directories: Disabled;
                    priority: 500;
   File inactivity threshold: 0;
File inactivity alert threshold: 0;
               remote access: *@pollux, *@pollux.sfb, *@oberon1,
                              *@oberon1.sfb, *@oberon2, *@oberon2.sfb,
                              *@leda1, *@leda1.sfb;
                 remote user: ;
                    password: ;
              backup command: ;
     application information: ;
                        ndmp: No;
             NDMP array name: ;
       De-duplication backup: No;
         De-duplication node: ;
         Probe resource name: ;
              virtual client: No;
               physical host: ;
           Proxy backup type: ;
           Proxy backup host: ;
             executable path: ;
    server network interface: ;
                     aliases: leda2, leda2.sfb;
                  index path: ;
          owner notification: ;
                 parallelism: 4;
               archive users: ;
               storage nodes: leda2;
       recover storage nodes: ;
         clone storage nodes: ;
                  hard links: Disabled;
             short filenames: Disabled;
                         BMR: Disabled;
                 BMR options: ;
                 backup type: ;
              client OS type: HpOS;
                        CPUs: 2;
           NetWorker version: LNMs_2007.Build.135;
              enabler in use: Yes;
       licensed applications: "NetWorker Module for Oracle, Unix Client/1";

Hi BR,

The HP-UX cluster is an active/active cluster. I think what you're saying is we use two more licenses than we actually need. My question is does this have any impact on functionality the way things are configured now? Why does one cluster have these problems and another doesn't? What is a curphyhost?

Martijn

Hi,

If active/active is used than you need two licenses each for one active node.

But you have to do the backup trough cluster host.

If you don't than index will be written under the physical host, configuration with DSN for LAN free will not work.

Configure new client for each cluster host on NW.

curphyhost is short for current physical host.

In that way (if you put curphyhost)when failover occurs for cluster host from one physical host to the other your LAN free backup will work without intervention.

BR

Hi BR,

You are right, this is how we configured all the Windows clusters in Networker as well. However I cannot change the current setup for the HP-UX cluster. What I do find puzzling is that it has worked like this in the past. It is still working on 1 HP-UX cluster, but for some reason it isn't working on the HP-UX cluster concerned anymore. Again thanks for your suggestions.

Martijn