rman backup restore questions and problems

From: Tilwen <tilwenbr_at_netscape.net>
Date: 26 Nov 2002 17:55:27 -0800
Message-ID: <8a41a7df.0211261755.45874920@posting.google.com>

I have a database about 100 gb in size. I do a full database backup every night. The other day, we had some corruption after a crash and it took a long time to do a restore on a tablespace, say 12 hours. I looked this up on metalink and there is a bug about the CKP table being over 10,000 records. They sent us a script to clean up this table. Well, that didn't help. I'm doing a test restore right now and it's going to take just as long to do it. The tablespace is only about 8gb in size. here is the commands we use to back it up. run
{ #ENV is case sensitive
  allocate channel t1 type 'sbt_tape' parms

'ENV=(DSMO_NODE=rx_r50,

              DSMI_CONFIG=/usr/tivoli/tsm/client/ba/bin/dsm.opt,
              DSMO_PSWDPATH=/usr/tivoli/tsm/client/oracle/bin,
              DSMI_DIR=/usr/tivoli/tsm/client/ba/bin,
              DSMO_AVG_SIZE=1000)';
  allocate channel t2 type 'sbt_tape' parms


        'ENV=(DSMO_NODE=rx_r50,

              DSMI_CONFIG=/usr/tivoli/tsm/client/ba/bin/dsm.opt,
              DSMO_PSWDPATH=/usr/tivoli/tsm/client/oracle/bin,
              DSMI_DIR=/usr/tivoli/tsm/client/ba/bin,
              DSMO_AVG_SIZE=1000)';
  allocate channel t3 type 'sbt_tape' parms


        'ENV=(DSMO_NODE=rx_r50,

              DSMI_CONFIG=/usr/tivoli/tsm/client/ba/bin/dsm.opt,
              DSMO_PSWDPATH=/usr/tivoli/tsm/client/oracle/bin,
              DSMI_DIR=/usr/tivoli/tsm/client/ba/bin,
              DSMO_AVG_SIZE=1000)';
 allocate channel t4 type 'sbt_tape' parms


        'ENV=(DSMO_NODE=rx_r50,

              DSMI_CONFIG=/usr/tivoli/tsm/client/ba/bin/dsm.opt,      
        DSMO_PSWDPATH=/usr/tivoli/tsm/client/oracle/bin,
              DSMI_DIR=/usr/tivoli/tsm/client/ba/bin,
              DSMO_AVG_SIZE=1000)';

  backup database;
  sql "alter system archive log current";   backup archivelog all delete input;

  release channel t1;
  release channel t2;
  release channel t3;
  release channel t4;

}
here is the restore script.
run
{ #ENV is case sensitive
  allocate channel t1 type 'sbt_tape' parms

'ENV=(DSMO_NODE=rx_r50,

              DSMI_CONFIG=/usr/tivoli/tsm/client/ba/bin/dsm.opt,
              DSMO_PSWDPATH=/usr/tivoli/tsm/client/oracle/bin,
              DSMI_DIR=/usr/tivoli/tsm/client/ba/bin,
              DSMO_AVG_SIZE=1000)';

  sql "alter tablespace d_cqm offline immediate";   restore tablespace d_cqm;
  recover tablespace d_cqm;
  sql "alter tablespace d_cqm online";
  release channel t1;
}
i do realize that I should being doing the restore in parallel also, but I don't think that that would cut down the restore into a 3rd. it might, but I doubt it.

I do have an idea but I want to run this by some of you. we are using adsm in conjuction with rman. Would it be better to put filesize paramater in the backup script. So that when doing the restore, rman doesn't have to read through so much of the backupset to get only a small portion of the backupset? He only has to read more backupsets, but they are a lot smaller(5 to 7 gb compared to 15 and 20 gb), so rman should read them faster.

I have done a search on here and looked on metalink, and came up with no leads. I'm posting here to get some insite maybe. I'm actually thinking that I will have to do a lot of testing to get something like this right. Yes, we should have been doing test restores a along to catch this. I know this, and have said this same thing before. We finally have gotten enough hardware and such to create a test DB, so things like this can be resolved before they are a production issue.

thanks. Received on Tue Nov 26 2002 - 19:55:27 CST