RAC: How to delete and add a node in 19c RAC

RAC: How to delete and add a node in 19c RAC (Doc ID 3019970.1)	​编辑To Bottom

---

In this Document	Goal

Solution

---

Applies to:

Oracle Database Enterprise Edition Version 19.21及以上版本

Goal

1. When a node is accessible, what steps should one take to remove it?
2. What can be done if a node cannot be accessed for deletion?
3. What actions are involved in manual removal of a node?
4. What process is involved in adding a new node?
5. What troubleshooting steps are necessary when adding or removing nodes?

Solution

NOTE: In the images shown below and in the following documents, user details such as cluster names, hostnames, directory paths, filenames, etc., stand for a fictional example (provided solely for illustrative purposes). There is no unintended resemblance to actual individuals or entities living or dead.

1. How to delete a node when the node is reachable?

By a root user, carry out the subsequent command on the node to be deleted:

cd $GI_home/crs/install

./rootcrs.sh -deconfig -force -verbose <<<<Make sure this is successful

From GI owner, go to $GI_home/deinstall and execute :

./deinstall -local <<<<Follow the onscreen instruction

1.2. Remove/Rename the following files and folder:

rm -rf $GRID_BASE <<<<Backup not needed

rm -rf /etc/oracle <<<<Backup not needed

rm -rf /var/tmp/.oracle <<<<Backup not needed

rm -f /etc/oraInst.loc <<<<Backup not needed

rm -f ora_inventory_location <<<<Backup needed

rm -f /etc/oratab <<<<Backup needed

1.3. Login to any working node and delete the node to be deleted as below:

crsctl delete node -n The node name can be identified by running the 'olsnodes' command-line tool.

1.4 The VIP of the node which was deleted as a root user was removed from the working node, and perform the necessary operations.

配置停止服务：<vip_name>此标记提示：通过执行以下命令可获取VIP名称：srvctl config nodeapps -a`

Srvctl remove vip -vip <vip_name>

1.5 Validate if the Node removed successfully or not as GI user, execute:

$GI_home/cluvfy stage -post nodedel -n

If the above command does not result in any error messages, the successful node deletion is achieved.

2. How to delete a node when the node is not reachable?

2.1.Login to the working node and execute as GI user:

cd $GI_home/oui/bin

./runInstaller --update-node-list ORACLE_HOME= "CLUSTER_NODES={}" -local

2.2. Login to any working node and delete the node to be deleted as below:

crsctl delete node -n > > > > Node name is retrieved via the olsnodes command.

Upon visiting, delete the VIP of that node which was previously deleted as a root user, then proceed to execute.

shut down the VIP service instance <vip_name> and retrieve its name via the command line interface: srvctl config nodeapps -a

srvctl remove vip -vip <vip_name>

2.4. Validate if the Node removed successfully or not as GI user, execute:

$GRID_BASE/cluvfy stage -post nodedel -n

Only if the command executed in the previous step does not result in an error message being returned, will the node deletion be considered successful.

3. How to delete a node manually?

3.1. Stop the CRS on the node which is to be deleted as Root, execute:

$GRID_BASE/bin/crsctl stop crs -f

假如某个原因导致CRS无法停止运行，则需要禁用其自启动功能并将其节点作为Root用户重启，请执行：

$GRID_BASE/bin/crsctl disable crs

3.2. Login to the working node and execute as GI user:

cd $GI_home/oui/bin

完整指定GI目录路径

运行安装程序并更新节点列表后，请指定ORACLE_HOME环境变量为【完整指定GI目录路径

3.3. Remove/Rename the following files and folder:

rm -rf $GRID_BASE <<<<Backup not needed

rm -rf /etc/oracle <<<<Backup not needed

rm -rf /var/tmp/.oracle <<<<Backup not needed

rm -f /etc/oraInst.loc <<<<Backup not needed

rm -f ora_inventory_location <<<<Backup needed

rm -f /etc/oratab <<<<Backup needed

3.4. Login to any working node and delete the node to be deleted as below:

Use the crsctl command to remove a node with the name <nodename> using the -n flag. Note that the node's name is accessible via the olsnodes command.

3.5. On the working node, remove the VIP of the node that was deleted as a Root user, execute the following steps.

The command srvctl stop vip -vip <vip_name> can be retrieved via executing the command srvctl config nodeapps -a.

srvctl remove vip -vip <vip_name>

3.6. Validate if the Node removed successfully or not as GI user, execute:

$GRID_BASE/cluvfy stage -post nodedel -n

Only if no error messages are returned upon execution of the above command will the node deletion be successful.

4. How to add a node in RAC cluster?

4.1. Invoke the "cluvfy" tool for pre-add node validation, where each working node is a Generalized Involved Participant (GI) owner, to perform the necessary validation tasks:

$GRID_BASE/cluvfy stage --prefix nodeadd --n --verbose <<<<Ensure this operation completes successfully

4.2. Execute the following from GI owner on any working node:

cd $GRID_BASE/addnode

注

At the conclusion of the aforementioned steps, it is directed to run root.sh on the remote node assuming the role of a Root user thereupon executing:

$GRID_BASE/root.sh <<<<Check the log file and make sure this is successful.

If the above command does not result in any error messages, the insertion of a node is successful.

5. Troubleshooting during Node addition or deletion.

By the deletion of a node, log files will be generated within $GRID_BASE/crsdata//crsconfig/

5.2.Inspect the cluster alert logs within $GRID_BASE/diag/crs//crs/trace.alert.log. The following information will include details about the system’s startup and shutdown events.

5.3. Check the deinstall logfile when executing ./deinstall -local

5.4. While performing node additions, log files will be generated within the /logs folder on the specific node where the 'nodeadd' command is initiated.

Upon execution of root.sh on a remote node, log files will be generated within the cluster directory located at $GRID_BASE/crsdata//crsconfig.

5.6. To check the VIP and Hostname, the following command can be executed:

hostname <<<<To get the public hostname

srvctl config nodeapps -a <<<<To get the VIP name

cat /etc/hosts <<<<To get the DNS names

||