ODAVP: Re-IP

This note is about changing the IP/netmask and gateway settings for a virtualized ODA.

These steps are detailed in the MOS 1504734.1 [How to Change Oracle Database Appliance IP Addresses Post Deployment], but mostly focusing on baremetal ODA.

 

 

#1 Boot the dom0 nodes

 

 

#2 Login to the first dom0 and reconfigure the 1 and optionally additional network

oakcli configure firstnet
oakcli configure additionalnet

If the second command show “no more interfaces to configure”, manually change the file /opt/oracle/oak/conf/dom0.xml

 

#3 Login with ‘grid’ to each dom0 systems and change the network settings in the clusterware, for example:

oifcfg delif -global eth1/128.3.10.0                <= old
oifcfg setif -global eth1/100.188.10.0:public <= new
oifcfg getif

 

#4 Login with root from the first dom0 node and bounce the clusterware to check for anomalies

crsctl stop cluster -all
crsctl start cluster -all

 

#5 Login with root from the first dom0 node and update the onecommand config file with the new parameter settings

/opt/oracle/oak/onecmd/onecommand.params 

 

#6 Stop the clusterware again

crsctl stop cluster -all

 

#7 Login with root from the first dom0 node and run through the wizatd, to check for anomalies

cd /opt/oracle/oak/onecmd
oakcli deploy -conf onecommand.params config

Warning this run with X, so it may be necessary to establish first, manually, the IP configuration of the eth1 network on Dom0

Make sure the root password does NOT include special characters (& or %)

Save the config at the end and compare with the previous version.

 

#8 Still from the first node, reconfigure the cluster with the new parameters

Run all steps that apply for the change

# cd /opt/oracle/oak/onecmd/
# ./GridInst.pl -o -l      <= list the step
# ./GridInst.pl -o -s 1     <= example for step1, here for Setup Network
# ./GridInst.pl -o -s 6     <= example for step6, here for Setup /etc/hosts

Check that the ifcfg-ethX files have been changed on both nodes

 

#9 Restart the clusterware, still from the first node 

# crsctl start cluster -all

 

#10 To change the VIP settings, as it turns that the above guideline could not complete this task, from each node:

# srvctl stop nodeapps -f
# srvctl modify vip -n <node> -netnum <num> -address <vip>/<gateway>/<interface>

 

#11 To change the SCAN VIP settings:

Check the SCAN settings with the following commands:

# srvctl config scan

If these settings are incorrect (ORA-12543 would eventually be returned from various TNS commands), then the scan vip need to be updated as well:

# srvctl stop scan_listener
# srvctl stop scan
# srvctl modify scan -n <scan>

Check that the changed has completed with the following command:

# srvctl config scan

 

This note is about changing the IP/netmask and gateway settings for a virtualized ODA.

These steps are detailed in the MOS 1504734.1 [How to Change Oracle Database Appliance IP Addresses Post Deployment], but mostly focusing on baremetal ODA.

 

#1 Boot the dom0 nodes

 

#2 Login to the first dom0 and reconfigure the 1 and optionally additional network

oakcli configure firstnet
oakcli configure additionalnet

If the second command show “no more interfaces to configure”, manually change the file /opt/oracle/oak/conf/dom0.xml

 

#3 Login with ‘grid’ to each dom0 systems and change the network settings in the clusterware, for example:

oifcfg delif -global eth1/128.3.10.0                <= old
oifcfg setif -global eth1/100.188.10.0:public <= new
oifcfg getif

 

#4 Login with root from the first dom0 node and bounce the clusterware to check for anomalies

crsctl stop cluster -all
crsctl start cluster -all

 

#5 Login with root from the first dom0 node and update the onecommand config file with the new parameter settings

/opt/oracle/oak/onecmd/onecommand.params 

 

#6 Stop the clusterware again

crsctl stop cluster -all

 

#7 Login with root from the first dom0 node and run through the wizatd, to check for anomalies

cd /opt/oracle/oak/onecmd
oakcli deploy -conf onecommand.params config

Warning this run with X, so it may be necessary to establish first, manually, the IP configuration of the eth1 network on Dom0

Make sure the root password does NOT include special characters (& or %)

Save the config at the end and compare with the previous version.

 

#8 Still from the first node, reconfigure the cluster with the new parameters

Run all steps that apply for the change

# cd /opt/oracle/oak/onecmd/
# ./GridInst.pl -o -l      <= list the step
# ./GridInst.pl -o -s 1     <= example for step1, here for Setup Network
# ./GridInst.pl -o -s 6     <= example for step6, here for Setup /etc/hosts

Check that the ifcfg-ethX files have been changed on both nodes

 

#9 Restart the clusterware, still from the first node 

# crsctl start cluster -all

 

#10 To change the VIP settings, as it turns that the above guideline could not complete this task, from each node:

# srvctl stop nodeapps -f
# srvctl modify vip -n <node> -netnum <num> -address <vip>/<gateway>/<interface>

 

#11 To change the SCAN VIP settings:

Check the SCAN settings with the following commands:

# srvctl config scan

If these settings are incorrect (ORA-12543 would eventually be returned from various TNS commands), then the scan vip need to be updated as well:

# srvctl stop scan_listener
# srvctl stop scan
# srvctl modify scan -n <scan>

Check that the changed has completed with the following command:

# srvctl config scan