Upgrade 10gR1 to 10gR2

This page describes the steps to upgrade PDB RACs 10gR1 + ASM from 10.1.0.4 to 10.2.0.2

• Perform a full backup of the DB before starting

upgrade from crs=10.1.0.4

• Backup OCR config and inventory before cleanup
  • save output of crsstat.sh, srvctl config service -d DBNAME; srvconfig -exp ...
  • DB create pfile='$HOME/work/init[DBNAME]_bck.ora' from spfile;
  • ASM create pfile='$HOME/work/initASM_bck.ora' from spfile;
  • save oraInventory/ContentsXML/inventory.xml
• stop db, asm, nodeapps and clusterware

  * edit/etc/inittab and telinit q
rm -rf /etc/init.d/init.crsd /etc/init.d/init.crs /etc/init.d/init.cssd /etc/init.d/init.evmd
rm -rf /etc/rc2.d/K96init.crs /etc/rc2.d/S96init.crs /etc/rc3.d/K96init.crs /etc/rc3.d/S96init.crs /etc/rc5.d/K96init.crs /etc/rc5.d/S96init.crs 
rm -rf /etc/oracle
rm -rf /var/tmp/.oracle 
rm -rf /etc/ORCLcluster
(optional) rm /etc/inittab.*

• create oraInventory.tar and rm -r oraInventory
• cleanup /etc/oratab,
• /etc/init.d/oma stop or /ORA/dbs01/oracle/product/10.2.0/agent10g/bin/emctl stop agent
• edit .bashrc (CRS_ORA_HOME, etc)

• map the new rawdevices for OCR and voting disk on one node
• copy devlabel config to other nodes and run devlabel reload
• cleanup raw devices with dd (make sure you have a procedure to restore services! and don't clean /dev/raw/raw3)
• chown oracle:ci /dev/raw/raw[123] /dev/raw/raw1[12] /dev/raw/raw22
• if needed, sudo ifconfig eth0:1 down
• Install from clusterware CD
  • ORA_CRS_HOME=/ORA/dbs01/oracle/product/10.2.0/crs
  • configure cluster and node names
    • cluster name can be name of the db
    • node names (public and private) -> edit as in the /etc/hosts file
  • 6 raw devices are needed:
    • 2 copies of OCR (/dev/raw/raw1, /dev/raw/raw11) take care of using 2 different disk arrays
    • 3 copies of the voting disk (/dev/raw/raw2, /dev/raw/raw12, /dev/raw/raw22) on 3 different disk arrays
    • /dev/raw/raw3 for the ASM spfile (don't need to modify)
• crsctl stop crs
• patch to 10.2.0.2
• go to upgrade DB

upgrade from crs=10.2.0.2

• dump info on services crsstat.sh or srvctl (use the versions from $ORA_CRS_HOME)
• create copy of pfile
  • create pfile='$HOME/init[DBNAME]_bck.ora' from spfile;
• stop the database, ASM and nodeapps
  • srvctl stop database -d [DBNAME] ,etc
• delete services from CRS
  • srvctl remove database -d [DBNAME]
  • srvctl remove asm -n ...
• use netca to remove LISTNER
• detach old oracle home

cd $ORACLE_HOME/oui/bin
./runInstaller -silent -removeHome ORACLE_HOME_NAME="OraRDBMSHOME" ORACLE_HOME="/ORA/dbs01/oracle/product/10.1.0/rdbms"

upgrade DB starts here from db=10.1.0.4 to crs=db=10.2.0.2

• copy tar of new oracle home
• if needed, edit bash with new value for ORACLE_HOME, etc
  • (logoff and logon to have also $PATH set easly and correctly)
• attach new Oracle Home (see steps here below)
• distribute the tar (Ex: scp rdbms_10_2_0_2.tar srv2:$PWD) at the destinations
  • • tar xfp rdbms_10_2_0_2.tar
  • find the appropriate value for ORACLE_HOME and its path
    • ex: see existing nodes: cat /ORA/dbs01/oracle/oraInventory/ContentsXML/inventory.xml
  • on the new nodes perform the clonig operation
    • cd $ORACLE_HOME/clone/bin
    • perl clone.pl ORACLE_HOME="/ORA/dbs01/oracle/product/10.2.0/rdbms" ORACLE_HOME_NAME="OraDb10g_rdbms" '-O"CLUSTER_NODES={[node1],[node2],}"' '-O"LOCAL_NODE=[nodename]"' (edit node names)
    • repeat for all new nodes, editing LOCAL_NODE value
    • run root.sh on new nodes, as instructed by clone.pl
• run netca to create listeners and register them with crs
  • remove extproc config
• add pfile and and password file for asm and db instances (on all nodes)
  • cp /ORA/dbs01/oracle/product/10.1.0/rdbms/dbs/init*.ora /ORA/dbs01/oracle/product/10.2.0/rdbms/dbs
  • orapwd file=./orapw+ASM1.. password=xxxxxx entries=20
  • orapwd file=./orapwdbinstance1.. password=xxxxxx entries=20

• check asm spfile
  • create pfile='/tmp/asmpfile13_03_pre.txt' from spfile='/dev/raw/raw3';
• change pfile and recreate spfile according to standards in installation guide and restart
• add asm instances and start them
  • srvctl add asm -n ... -i +ASM1 -o $ORACLE_HOME
• cp /ORA/dbs01/oracle/product/10.1.0/rdbms/network/admin/tnsnames.ora /ORA/dbs01/oracle/product/10.2.0/rdbms/network/admin/
• edit database pfile following new standard (see installation guide)
  • edit _recyclebin, cluster_database and other params
  • edit shared_pool (for example with __shared_pool_size params) to be >=500MB (??)
• cd $ORACLE_HOME/rdbms/admin
• startup upgrade pfile=$HOME/init[DBNAME]_migr.ora (the edited pfile)
• SPOOL upgrade.log
• DO YOU HAVE BACKUPS ?
• @catupgrd.sql
• shutdown and startup pfile=$HOME/init[DBNAME]_migr.ora (still cluster=false)
• run @utlrp.sql
• shutdown, edit pfile to set cluster database=true and
  • create create spfile='+DATA_DG1/DBNAME/spfileDBNAME.ora' from pfile='.. ';
  • check and if needed update $ORACLE_HOME/dbs/initdbinst.ora (all nodes)
• add db to CRS

srvctl add database -d DBNAME-o $ORACLE_HOME -p '+DATA_DG1/DBNAME/spfileDBNAME'
srvctl add instance -d DBNAME -i INSTNAME -n ...
srvctl add instance -d DBNAME -i INSTNAME -n ...

srvctl modify instance -d DBNAME -i INSTNAME -s +ASM... # add dependencies between db instance and asm

• startup using srvctl
• recreate services

srvctl add service -d ... -s .... -r "instname1,instaname2,..."
if needed, srvctl start service -d ..

• edit /etc/logrotate.d/ora_cern_listener_udump_bdump_adump_rotate
• on all nodes of the cluster grant the 'sysdba' privilege to superusers.
• update /etc/oratab files on all nodes of the cluster
• install or reattach the Oracle_HOME for the EM agent

• Perform a full backup of the DB when finished