Due to complex requirements for service relocate and restore as part of patching, I decided to script the process.
Requirements: ```code Critical services should only be relocated once as part of patching. After patching, non-critical services will need to be relocated to other instance. Critical services are defined as svc_21-29.
## Initial Environment and Service Configuration Existing configuration: ```code [oracle@racnode-dc2-1 patch]$ . /media/patch/hawk.env The Oracle base has been set to /u01/app/oracle ORACLE_UNQNAME=hawk ORACLE_SID=hawk1 ORACLE_BASE=/u01/app/oracle ORACLE_HOME=/u01/app/oracle/12.2.0.1/db1 Oracle Instance alive for sid "hawk1" [oracle@racnode-dc2-1 patch]$ srvctl status database -d $ORACLE_UNQNAME -v Instance hawk1 is running on node racnode-dc2-1 with online services svc_11,svc_12,svc_13,svc_14,svc_15,svc_16,svc_17,svc_18,svc_19. Instance status: Open. Instance hawk2 is running on node racnode-dc2-2 with online services svc_21,svc_22,svc_23,svc_24,svc_25,svc_26,svc_27,svc_28,svc_29. Instance status: Open.
Capturing Baseline Service Affinity
Save existing services affinity:
[oracle@racnode-dc2-1 patch]$ ./save_service.sh $ORACLE_UNQNAME ##### Save services affinity: /tmp/service_hawk1.conf Instance hawk1 is running on node racnode-dc2-1 with online services svc_11,svc_12,svc_13,svc_14,svc_15,svc_16,svc_17,svc_18,svc_19. Instance status: Open. ##### Save services affinity: /tmp/service_hawk2.conf Instance hawk2 is running on node racnode-dc2-2 with online services svc_21,svc_22,svc_23,svc_24,svc_25,svc_26,svc_27,svc_28,svc_29. Instance status: Open. -r-------- 1 oracle oinstall 156 Jun 3 17:40 /tmp/service_hawk1.conf -r-------- 1 oracle oinstall 156 Jun 3 17:40 /tmp/service_hawk2.conf
Note: /tmp/service_hawk*.conf is read-only to prevent accidental overwrite.
Relocating Services for Instance-Level Patching
Patching starts from instance1 and relocate all services to instance2.
[oracle@racnode-dc2-1 patch]$ ./relocate_service.sh $ORACLE_UNQNAME 1 2 ******************************** ***** SERVICES CURRENT LOCATION: ******************************** Instance hawk1 is running on node racnode-dc2-1 with online services svc_11,svc_12,svc_13,svc_14,svc_15,svc_16,svc_17,svc_18,svc_19. Instance status: Open. Instance hawk2 is running on node racnode-dc2-2 with online services svc_21,svc_22,svc_23,svc_24,svc_25,svc_26,svc_27,svc_28,svc_29. Instance status: Open. ****************************** ***** CMD RELOCATE SERVICES TO: hawk2 ****************************** + srvctl relocate service -d hawk -service svc_11 -oldinst hawk1 -newinst hawk2 ... Instance hawk1 is running on node racnode-dc2-1. Instance status: Open. Instance hawk2 is running on node racnode-dc2-2 with online services svc_11,svc_12,svc_13,svc_14,svc_15,svc_16,svc_17,svc_18,svc_19,svc_21,svc_22,svc_23,svc_24,svc_25,svc_26,svc_27,svc_28,svc_29. Instance status: Open.
Patching completed from instance1. Patching starts from instance2 and relocate all services to instance1.
[oracle@racnode-dc2-1 patch]$ ./relocate_service.sh $ORACLE_UNQNAME 2 1 ******************************** ***** SERVICES CURRENT LOCATION: ******************************** ... ****************************** ***** CMD RELOCATE SERVICES TO: hawk1 ****************************** + srvctl relocate service -d hawk -service svc_11 -oldinst hawk2 -newinst hawk1 ... + srvctl relocate service -d hawk -service svc_21 -oldinst hawk2 -newinst hawk1 ... Instance hawk1 is running on node racnode-dc2-1 with online services svc_11,svc_12,svc_13,svc_14,svc_15,svc_16,svc_17,svc_18,svc_19,svc_21,svc_22,svc_23,svc_24,svc_25,svc_26,svc_27,svc_28,svc_29. Instance status: Open. Instance hawk2 is running on node racnode-dc2-2. Instance status: Open.
Finalizing Service Distribution and Restoring Affinity
Patching is completed and all services are running from instance1. Instead of relocating services back to its original instance, non-critical services are relocated to the other instance. Essentially, svc_11-19 is relocated 3x while svc_21-29 is relocated 1x.
Non-Critical services were running from instance1 per /tmp/service_hawk1.conf when save_service.sh was run. Non-Critical services are currently running from instance1 and will need to be restored to instance2.
[oracle@racnode-dc2-1 patch]$ ./restore_service.sh $ORACLE_UNQNAME 1 2 /tmp/service_hawk1.conf ******************************** ***** SERVICES CURRENT LOCATION: ******************************** Instance hawk1 is running on node racnode-dc2-1 with online services svc_11,svc_12,svc_13,svc_14,svc_15,svc_16,svc_17,svc_18,svc_19,svc_21,svc_22,svc_23,svc_24,svc_25,svc_26,svc_27,svc_28,svc_29. Instance status: Open. Instance hawk2 is running on node racnode-dc2-2. Instance status: Open. ************************** ***** SAVED CONFIGURATION: /tmp/service_hawk1.conf ************************** Instance hawk1 is running on node racnode-dc2-1 with online services svc_11,svc_12,svc_13,svc_14,svc_15,svc_16,svc_17,svc_18,svc_19. Instance status: Open. ************************************* ***** CMD - RESTORE SERVICES TO: hawk2 ************************************* + srvctl relocate service -d hawk -service svc_11 -oldinst hawk1 -newinst hawk2 ... Instance hawk1 is running on node racnode-dc2-1 with online services svc_21,svc_22,svc_23,svc_24,svc_25,svc_26,svc_27,svc_28,svc_29. Instance status: Open. Instance hawk2 is running on node racnode-dc2-2 with online services svc_11,svc_12,svc_13,svc_14,svc_15,svc_16,svc_17,svc_18,svc_19. Instance status: Open.
Automation Scripts: Save, Relocate, and Restore
Hopefully, patching is not always this complex and if it is, maybe the scripts will be useful.
save_service.sh
#!/bin/sh -e # MDinh : Mar 27, 2019 # DN=`dirname $0` BN=`basename $0` DB=${1:?"---> USAGE: $DN/$BN <db_unique_name>"} IFS="," save_service () { OUTF=/tmp/service_$i.conf srvctl status instance -d ${DB} -instance ${i} -v > $OUTF echo echo "##### Save services affinity: $OUTF" chmod 400 $OUTF cat $OUTF echo } inst=$(srvctl config database -db ${DB} |grep "^Database instances" | awk -F' ' '{print $3}') for i in ${inst}; do save_service done ls -lh /tmp/service*.conf exit
relocate_service.sh
#!/bin/sh # relocate_service.sh # MDinh : Jun 02, 2019 # DN=`dirname $0` BN=`basename $0` DB=${1:?"---> USAGE: $DN/$BN <db_unique_name> <oldinst> <newinst>"} OLD=${2:?"---> USAGE: $DN/$BN <db_unique_name> <oldinst> <newinst>"} NEW=${3:?"---> USAGE: $DN/$BN <db_unique_name> <oldinst> <newinst>"} echo echo "********************************" echo "***** SERVICES CURRENT LOCATION:" echo "********************************" srvctl status database -d ${DB} -v echo echo "******************************" echo "***** CMD RELOCATE SERVICES TO: ${DB}${NEW}" echo "******************************" OUTF=/tmp/old_service_${DB}${OLD}.loc srvctl status instance -d ${DB} -instance ${DB}${OLD} -v > $OUTF svc=$(tail -1 $OUTF| awk -F" " '{print $11}'| awk '{$0=substr($0,1,length($0)-1); print $0}') IFS="," for s in ${svc} do set -x srvctl relocate service -d ${DB} -service ${s} -oldinst ${DB}${OLD} -newinst ${DB}${NEW} set +x done srvctl status database -d ${DB} -v exit
restore_service.sh
#!/bin/sh # restore_service.sh # MDinh : Jun 02, 2019 # DN=`dirname $0` BN=`basename $0` DB=${1:?"---> USAGE: $DN/$BN <db_unique_name> <oldinst> <newinst> <conf>"} OLD=${2:?"---> USAGE: $DN/$BN <db_unique_name> <oldinst> <newinst> <conf>"} NEW=${3:?"---> USAGE: $DN/$BN <db_unique_name> <oldinst> <newinst> <conf>"} CONF=${4:?"---> USAGE: $DN/$BN <db_unique_name> <oldinst> <newinst> <conf>"} echo echo "********************************" echo "***** SERVICES CURRENT LOCATION:" echo "********************************" srvctl status database -d ${DB} -v echo echo "**************************" echo "***** SAVED CONFIGURATION: $CONF" echo "**************************" cat $CONF echo echo "*************************************" echo "***** CMD - RESTORE SERVICES TO: ${DB}${NEW}" echo "*************************************" svc=$(tail -1 $CONF| awk -F" " '{print $11}'| awk '{$0=substr($0,1,length($0)-1); print $0}') IFS="," for s in ${svc} do set -x srvctl relocate service -d ${DB} -service ${s} -oldinst ${DB}${OLD} -newinst ${DB}${NEW} set +x done srvctl status database -d ${DB} -v exit
Oracle Database Consulting Services
Ready to optimize your Oracle Database for the future?
.svg){kind=small}
.svg){kind=small}