This article contains only the FlashGrid-specific steps required for upgrading Oracle Grid Infrastructure from 19c to 26ai.
The overall upgrade should follow Oracle's Grid Infrastructure upgrade procedure , together with the FlashGrid-specific steps described in this article.
Note: FlashGrid software version 26.05 or later is required.
Pre-Upgrade Steps
1. Enable the FlashGrid Cloud Configuration Service
Run the following command on all database nodes as root:
systemctl enable --now flashgrid-os-conf-cloud.service
Verify:
systemctl status flashgrid-os-conf-cloud.service
2. Verify the Grid Infrastructure OSDBA Group
FlashGrid ASM devices use the asmadmin group. Verify the OSDBA group configured in the existing 19c Grid Infrastructure home.
export GRID_HOME=/u01/app/19.3.0/grid $GRID_HOME/bin/osdbagrp $GRID_HOME/bin/osdbagrp -a
Both commands should return:
asmadmin
If both commands return asmadmin, no changes are required. Skip Step 3 and continue with the remaining upgrade steps.
If the first command returns asmdba, complete Step 3 before starting the Grid Infrastructure upgrade.
3. Change the Grid Infrastructure OSDBA Group to asmadmin
Complete this step only if Step 2 shows that the existing Grid Infrastructure OSDBA group is configured as asmdba.
Update the existing Grid Infrastructure home one node at a time.
Stop CRS and unlock the GI home as root:
export GRID_HOME=/u01/app/19.3.0/grid $GRID_HOME/bin/crsctl stop crs cd $GRID_HOME/crs/install ./rootcrs.sh -unlock
As the grid user, update config.c:
su - grid export ORACLE_HOME=/u01/app/19.3.0/grid sed -i -E \ -e 's|^([[:space:]]*\.Ldba_string:[[:space:]]+\.string[[:space:]]+)"[^"]*"|\1"asmadmin"|' \ -e 's|^#define[[:space:]]+SS_DBA_GRP[[:space:]]+"[^"]*"|#define SS_DBA_GRP "asmadmin"|' \ $ORACLE_HOME/rdbms/lib/config.c
As the grid user, rebuild config.o:
cd $ORACLE_HOME/rdbms/lib rm -f config.o make -f ins_rdbms.mk \ $ORACLE_HOME/rdbms/lib/config.o
The following warning can be ignored:
cat: /etc/flashgrid-image-release: Permission denied
Rebuild osdbagrp:
cd $ORACLE_HOME/srvm/lib make -f ins_srvm.mk iosdbagrp
Verify the updated configuration:
$ORACLE_HOME/bin/osdbagrp $ORACLE_HOME/bin/osdbagrp -a
Both commands should now return:
asmadmin
Lock the GI home and start CRS again as root:
export GRID_HOME=/u01/app/19.3.0/grid cd $GRID_HOME/rdbms/install ./rootadd_rdbms.sh cd $GRID_HOME/crs/install ./rootcrs.sh -lock $GRID_HOME/bin/crsctl start crs -wait
Repeat the procedure on the remaining database nodes, one node at a time.
During-Upgrade Steps
4. Review FlashGrid-Related Prerequisite Warnings
Review the Oracle 26ai Grid Infrastructure prerequisite check results. The following environment-specific warnings may be ignored when applicable:
- Swap configuration
- NTP/Chrony external time source
5. Launch the Grid Infrastructure Upgrade
Run the Grid Infrastructure upgrade as the grid user with -ignorePrereqFailure:
./gridSetup.sh -silent -upgrade -ORACLE_BASE /u01/app/grid -ignorePrereqFailure
Continue the remaining steps according to Oracle's Grid Infrastructure upgrade procedure .
Post-Upgrade Steps
6. Update Grid User Environment Variables
Update the environment variables for the grid user on all database nodes.
Update /home/grid/.bashrc and/or /home/grid/.bash_profile:
export ORACLE_BASE=/u01/app/grid export ORACLE_HOME=/u01/app/23.0.0/grid export PATH=$ORACLE_HOME/bin:$PATH
Reload the environment:
source ~/.bash_profile
7. Update Root Environment and Scripts
If the root user or any administrative scripts reference the old Grid Infrastructure home, update:
/u01/app/19.3.0/grid
to:
/u01/app/23.0.0/grid
Search for references to the old GI home:
grep -Rni '/u01/app/19.3.0/grid' \ /root/.bash_profile \ /root/.bashrc \ /etc/profile.d 2>/dev/null
Update any applicable scripts or environment files.
8. Restart the FlashGrid ASM Service and Validate the Cluster
Restart the FlashGrid ASM service as root on all database nodes:
systemctl restart flashgrid_asm
Verify the FlashGrid cluster:
flashgrid-cluster
Confirm that the cluster, nodes, disks, and ASM disk groups are healthy. All disk groups must have zero offline disks and Resync = No before it is safe to proceed with the next node.