BigBlog

Success

This blog continues at the new site.

I am having problems with using the Data Guard wizard under Grid Control:

1. The cluster on botany was not recognised by srvctl.  Fixed - see below.
2. The monitoring user for a physical standby database must have SYSDBA (the user is hardcoded as DBSNMP in Grid Control)
3. The password for SYS is not recognised when I use the wizard to access the database cluster on botany.  Works for local access.

To fix the srvctl problem,
 srvctl add database -d BOTANY -o $ORACLE_HOME -r PHYSICAL_STANDBY -n example -m yaocm.id.au -p +DATA/botany/spfileexample.ora

• -d corresponds to the DB_UNIQUE_NAME
• -n corresponds to the database name

Once this was done, Grid Control recognised the cluster database once agent discovery was complete (executed on both nodes):
$ORA_AGENT_HOME/bin/agentca -d -c botany1,botany2

I have managed to duplicate the example database onto the botany cluster using RMAN.  The standby database has been registered to Grid Control.

However, I am having difficulty in getting Grid Control to manage the Standby configuration because of mismatched service names - the standby database has BOTANY.yaocm.id.au whereas Grid Control is expecting example2.yaocm.id.au.

The rough procedure is:

• Replicate the directory structure for $ORACLE_BASE/admin/example/[abcu]dump on both nodes
• Create the directory structure in ASM via ASMCMD mkdir command for:
• +DATA/example
• +DATA/BOTANY
• +FRA/example
• +FRA/BOTANY
• Create a directory called /u00/backup on padstow2 and botany2 (these nodes have more disk space because the installation software was not loaded there)
• Fiddle around with the listener configurations (need to clarify this in my next attempt)
• Using RMAN on padstow2,
• rman target / catalog rman@rmancat
• backup validate database; /* checks integrity of database */
• blockrecover corruption list;  /* repairs any damaged blocks */
• backup format='/u00/backup/%U.dbf' database;
• backup format='/u00/backup/%U.dbf' current controlfile for standby;
• sql 'alter system switch logfile';  /* also done on padstow1 */
• backup format='/u00/backup/%U.dbf' archivelog all; /* wrong order, but it works? */
• sql "create pfile='/u00/backup/initexample2.ora' from spfile";
• restore validate database; /* checks the integrity of the backup */
• exit
• scp /u00/backup/* botany2:/u00/backup/
• On botany2, update /etc/oratab with:
  example2:/u00/app/oracle/product/10.2.0/db_1:N
• export ORACLE_HOME=/u00/app/oracle/product/10.2.0/db_1
• export ORACLE_SID=example2
• export PATH=$ORACLE_HOME/bin:$PATH  # or I could use . oraenv
• vi /u00/backup/initexample2.ora to change the DB_UNIQUE_NAME to BOTANY
• vi $ORACLE_HOME/dbs/initexample2.ora to add:
  spfile='+DATA/example/spfileexample'
• sqlplus / as sysdba
• startup nomount pfile='/u00/backup/initexample2.ora'
• create spfile='+DATA/example/spfileexample' from pfile='/u00/backup/initexample2.ora';
• shutdown immediate
• exit
• rman target sys@padstow2 catalog rman@rmancat auxiliary /
• startup clone nomount
• duplicate target database for standby dorecover;

• ALTER DATABASE RECOVER MANAGED STANDBY DATABASE DISCONNECT FROM SESSION;

At this point, the standby database appears to be up and running.

To get Grid Control to recognise the new database, I had to open the database as read-only:

• sqlplus / as sysdba
• shutdown immediate
• startup mount
• alter database open read only;

Now the Grid Control is able to connect to the database and is thus able to be registered.

I am now longer primarily focussed on the OCM exam but am seeking work.

I have already had one (1) interview (thanks to Jeff R. for the lead).

After one (1) reschedule then cancellation of the OCM exam, Oracle University has now scheduled an exam for the 13 to 14 Nov 2008.  I have submitted an enrolment request (more fool me).

The Data Guard concepts are slowly making sense.  My three (3) attempts so far have failed but I am gaining experience in diagnosing failures in Data Guard set up.