Forum Discussion

dtrostli's avatar
dtrostli
Level 4
12 years ago
Solved

RMAN restore ends with error

Hello Everyone,

Has anyone come across an error while trying to restore using RMAN. The following detail is showing in the job log for the restore. The main issue is that the restore worked 2 out 12 tries. Any help is very much appreciated.

 

1/14/2014 1:59:57 PM - begin Restore

1/14/2014 1:59:59 PM - 1 images required

1/14/2014 1:59:59 PM - media MGU013 required

1/14/2014 2:00:03 PM - restoring image s5a9oc01_1389428357

1/14/2014 2:00:05 PM - Info bpbrm(pid=4816) s5a9octst01 is the host to restore to      

1/14/2014 2:00:05 PM - Info bpbrm(pid=4816) telling media manager to start restore on client    

1/14/2014 2:00:07 PM - Warning bptm(pid=312) failure logging message to client s5a9octst01 in log /usr/openv/netbackup/logs/user_ops/dbext/logs/17007.0.1389715202:  system error occurred (130)

1/14/2014 2:00:08 PM - requesting resource MGU013

1/14/2014 2:00:08 PM - granted resource MGU013

1/14/2014 2:00:08 PM - granted resource HP.ULTRIUM5-SCSI.000

1/14/2014 2:00:08 PM - mounted

1/14/2014 2:00:08 PM - Info bpbrm(pid=2740) s5a9octst01 is the host to restore to     

1/14/2014 2:00:08 PM - Info bptm(pid=312) MGU013           

1/14/2014 2:00:08 PM - Info bpbrm(pid=2740) listening for client connection        

1/14/2014 2:00:08 PM - Info bptm(pid=312) INF - Waiting for positioning of media id MGU013 on server s5a9nbu01 for reading.

1/14/2014 2:00:08 PM - positioning MGU013 to file 84

1/14/2014 2:00:09 PM - Error bpbrm(pid=2740) cannot become user oracle and group oinstall on s5a9octst01   

1/14/2014 2:00:09 PM - Error bpbrm(pid=2740) listen for client protocol error - couldn't write necessary information on /usr/openv/netbackup/logs/user_ops/dbext/logs/17007.0.1389715202

1/14/2014 2:00:10 PM - Info bpbrm(pid=4816) child done, status 25        

1/14/2014 2:00:10 PM - Info bpbrm(pid=4816) sending message to media manager: STOP RESTORE s5a9oc01_1389428357    

1/14/2014 2:00:41 PM - positioned MGU013; position time: 00:00:33

1/14/2014 2:00:41 PM - begin reading

1/14/2014 2:01:06 PM - Info bpbrm(pid=4816) media manager for backup id s5a9oc01_1389428357 exited with status 150: termination requested by administrator

 

Kind Regards


David

7.5
Backup and Recovery
Linux
NetBackup

19 Replies

Sort By
  • RiaanBadenhorst's avatar
    RiaanBadenhorst
    Level 6
    12 years ago

    Error bpbrm(pid=2740) cannot become user oracle and group oinstall on s5a9octst01

     

    Check this note. http://www.symantec.com/docs/TECH20582

  • dtrostli's avatar
    dtrostli
    Level 4
    12 years ago

    Hello Riaan,

     

    Customer calls the restore from the client. It does not have any script with su in it. It is a simple restore using rman commands. The client is running RedHat 6.3.


    Regards


    David

  • dtrostli's avatar
    dtrostli
    Level 4
    12 years ago

    Yes, They are. I have just seen too many CLOSE_WAIT and TIME_WAIT connections. Would that affects NetBackup communication? I know Symantec has some recommended tuning for it but I can't find the information. Do you by any chance have anything on TCP tuning?

  • RiaanBadenhorst's avatar
    RiaanBadenhorst
    Level 6
    12 years ago

    The failure occurs to quickly too be network related (IMHO). Has this worked before? Maybe post the script /commans they're using so it can be checked.

    EDIT, sorry saw you said it works 2/12. Its a bit weird then.

     

    Does seem somehow related to comms, or maybe the comms processes. Check this note (older and for backup but same symptoms)

     

    http://www.symantec.com/business/support/index?page=content&id=TECH77533

  • dtrostli's avatar
    dtrostli
    Level 4
    12 years ago

    Hi,

     

    This note is for older versions. This does not apply to 7.x. In 7.x bpcd and vnetd are standalone processes. I have seen in the past issues with too many CLOSE_WAIT and TIME_WAIT connections. But that was too long ago. Symantec has sent me some tunig for RedHat to tune the TCP layer. Hope that works otherwise customer will be in my throat. 

  • dtrostli's avatar
    dtrostli
    Level 4
    12 years ago

    Hi,

    If you find anything else related to my problem, please let me know.

    Thanks

  • Marianne's avatar
    Marianne
    Level 6
    12 years ago

    Firstly check that user oracle and group oinstall on s5a9octst01 have the same UID and GID as the source system.

    Next, ensure all of these log folders exist for further troubleshooting:

    On destination client: dbclient (with 777 permissions) and bpcd

    On master: bprd (if folder does not exist, create it and restart NBU to enable this log)

    On media server: bpbrm and bptm

    Copy latest logs to reflect process name (e.g. dbclient.txt) and upload as File attachment.

     

    One more thought:

    How many NICs/IPs on destination client?
    If more than one, is each IP address bound to a different hostname in its own hosts file?
     

  • dtrostli's avatar
    dtrostli
    Level 4
    12 years ago

    Hi Marianne,

    I have asked customer to check the oracle ID and group ID is the same on all of the servers. I have also create the logs with the correct permission. In the meantime I don't see why having a second nic would impact restore. All the communication flows through the production network for the test restore despite the server having a backup network as well. The weird thing is that restore worked twice out of 10 tries and I see in the dbclient a cannot connect on socket error. I am suspecting more of a socket error than anything else.

    David