NBU 5.0 for windows, oracle database backup failed with the ORA-27206 error

[Stanley_Seow]

Hi all,

I have 1 master server and 2 oracle servers, both backing up to the master server under a dedicatede backup LAN. All 3 are running Windows 2000. I've faced some problems with the Oracle agent.

Here's the log from RMAN:

RMAN-00571: ===========================================================
RMAN-00569: =============== ERROR MESSAGE STACK FOLLOWS ===============
RMAN-00571: ===========================================================
RMAN-03009: failure of backup command on ch00 channel at 03/10/2005 14:28:36
RMAN-10032: unhandled exception during execution of job step 1:
ORA-19583: conversation terminated due to error
ORA-27206: requested file not found in media management catalog
ORA-06512: at "SYS.DBMS_BACKUP_RESTORE", line 625
ORA-06512: at line 206
RMAN-10035: exception raised in RPC:
ORA-19583: conversation terminated due to error
ORA-27206: requested file not found in media management catalog
ORA-06512: at "SYS.DBMS_BACKUP_RESTORE", line 625
RMAN-10031: RPC Error: ORA-19583 occurred during call to DBMS_BACKUP_RESTORE.BA
CKUPPIECECREATE

Most of the suggestions in this forum and support.veritas.com are about this problem happening in a UNIX platform. But for me, it is happening on a Windows platform. Anyhow, here're the suggestions from support site and steps that i've taken
1. Checked through the server and client name and they have no problem.
2. Added the REQUIRED_INTERFACE entry in the NBU registry with the correct IP of the client.
3. Checked that there's no dbwr_io_slaves value in the init.ora file.
4. Run the exact same backup script in RMAN environment and it still return the same msg.

The same thing still happens, can anyone please help me here? Thank you.

[Stanley_Seow]

Latest update about the problem.:

1. Create a new oracle policy with the default application backup shcedule only.
2. Perform user backups from the client and everything works fine.
3. Create a backup schedule at the same oracle policy and set the template used for the manual backup at step 2, after that manual initiate the policy.

4. It fails with the exact same message previously.
5. Perform user backup from the client and it fails now with the same message.
6. So I suspect this is caused by the schedule backup and the backup template, I remove the template from the policy and perform the client backup, fails with the same msg.

7. I remove the schedule backup from the policy, so now there's only the default application backup schedule in the policy, I perform the client backup, fails again.

8. Remove the whole policy, recreate it with the exact same default application backup schedule and perform client backup. It works again.

So, we're sure that it's now the backup schedule and the template pointing problem? I suspect there's still some problem with my server/client naming mismatch, but why does it work with client initiated backup? I really have no idea where to fix now, please advise. Thank you.

[Brice_Avila_2]

Stanley,

ORA-27206 errors are mostly seen when these three references for the client's hostname do not match:

1. The client name in the Oracle policy
2. The CLIENT_NAME parameter in the client's "/usr/openv/netbackup/bp.conf" file
3. The $NB_ORA_CLIENT environmental variable

All three references must match each other. Examples of these differences include:

* Short name and fully qualified long name (i.e. "hostname" and "hostname.veritas.com")
* Case distinction between the hostnames (i.e. "hostname" and "HOSTNAME")
* Discontinued SERVER entries in the "/usr/openv/netbackup/bp.conf" files
* A hostname missing from the /etc/hosts file, or DNS
* If running multiple NICs, needing a REQUIRED_INTERFACE setting (page 158 of the NetBackup 5.1 System Administrator's Guide, Volume II for UNIX, available at "http://support.veritas.com/docs/268091")
* Using an IP address instead of a client's hostname

Sometimes explicitly specifying the policy, server and client in the RMAN script similar to the following has been known to fix the problem:

run {
allocate channel t1 type 'sbt_tape'
parms="ENV=(NB_ORA_POLICY= policy_name, NB_ORA_SERV= server_name, NB_ORA_CLIENT = client_name)";
allocate channel t2 type 'sbt_tape'
parms="ENV=( NB_ORA_POLICY= policy_name, NB_ORA_SERV= server_name, NB_ORA_CLIENT = client_name)";
backup
(database format 'bk_%s_%p_%t');
}

With VCS clustered clients, there may be the need to add some altname references for each client. For example, given two nodes in a VCS cluster ("node1", "node2") and a virtual name ("virt_name"):

1. In the Oracle policy specify the virtual name in the in the "clients" tab
2. Specify "CLIENT_NAME = virt_name" in the /usr/openv/netbackup/bp.conf file on the client.
3. On the master server:

mkdir /usr/openv/netbackup/db/alnames
echo "virt_name" > /usr/openv/netbackup/db/altnames/node1
echo "virt_name" > /usr/openv/netbackup/db/altnames/node2

The following bpclntcmd commands can help you confirm your server names. Please refer to page 37-39 of the NetBackup 5.1 Troubleshooting Guide at "http://support.veritas.com/docs/268196" for more information:

/usr/openv/netbackup/bin/bpclntcmd -pn
/usr/openv/netbackup/bin/bpclntcmd -hn
/usr/openv/netbackup/bin/bpclntcmd -ip
/usr/openv/netbackup/bin/bpclntcmd -hn
/usr/openv/netbackup/bin/bpclntcmd -ip < IP address of client>

Below is further information that may assist:

http://support.veritas.com/docs/233512
http://support.veritas.com/docs/252339

Hope this helps. Good luck.

Brice Avila