cancel
Showing results for 
Search instead for 
Did you mean: 

Backup many oracle db on single host failure

xmenvox
Level 2

Hi all,

We have many policy backup for many oracle db 11g and 12c (about 6 dbs) on two node cluster AIX 7.1. Backup job frequently fail with some errors:

1. the backup failed to back up the requested files  (6)

2. Database is in the wrong state (must be OPEN) for the requested action  (5407)

3. cannot connect on socket  (25)

Master server Windows, media server Redhat 7.7.6

Have any idea for this case.

Thank you so much.

3 REPLIES 3

Marianne
Moderator
Moderator
Partner    VIP    Accredited Certified
No 1 could be caused by no 2.
Your dba's need to look at that.

3. (Status 25) is a connectivity issue. You need to check the source and destination ip addresses of the failed connection attempt and ensure port 1556 is open in both directions.

Tousif
Level 6

Hello,

 

1. the backup failed to back up the requested files  (6)

-> This very generic error for oracle. If any child job get fail the parent one fail with status 6.

2. Database is in the wrong state (must be OPEN) for the requested action  (5407)

-> You can not take oracle database backup using oracle policy type if database is shutdown, nomount or mounted state. The database must have open state. You can check with DBA about the database state.

3. cannot connect on socket  (25)

-> Its NBU error for the client not able to communicate on mentioned client name or ip. To test the communication run the below command from master.

<InstallPath>\NetBackup\bin\admincmd\bptestbpcd -client <ClientName> -verbose -debug

Review the command output and check if it resolving the correct IP and hostname.

Regards,

Marianne
Moderator
Moderator
Partner    VIP    Accredited Certified

@xmenvox

Some additional thoughts:

Status 6 is a generic error saying that the rman script has failed. 
Your dba's need to look at the rman output to find the cause.
No. 2 is self-explanatory. If dba's are doing some sort of maintenance, they need to inform you to disable the backup policy.

3. (Status 25) is a connectivity issue. You need to check the source and destination ip addresses of the failed connection attempt and ensure port 1556 is open in both directions.

Something else to double-check is the hostname used in the backup policy.
If this is a normal failover cluster, the virtual hostname for the instance must be used. 
There should also be a NB_ORA_CLIENT parameter in the rman script with virtual hostname. 
Double-check forward and reverse name lookup between master and media server(s) and virtual hostname.
Test connectivity to virtual hostname using bptestbpcd.

If this is a RAC environment, you need to find out how the cluster is configured and set up policies/script(s) accordingly.
Read through the RAC chapter in NBU for Oracle manual and work with your dba.