Scaling Up Master Node Specifications in an MRS Cluster

As service workloads grow and Core nodes are added, cluster CPU utilization increases. The original Master node specifications may no longer meet service requirements. Insufficient Master node resources degrade cluster management performance and component stability. Upgrade the Master node specifications to maintain normal cluster operations. You can scale up Master node specifications on the MRS console using either of the following methods: One-Click Upgrade or Step-by-Step Upgrade.

Prerequisites

If Host Security Service (HSS) is enabled on the cluster, disable HSS monitoring before upgrading Master node specifications. After the upgrade completes, re-enable HSS monitoring.
Ensure that sufficient specification resources are available throughout the steps in Scaling Up Master Node Specifications (Step-by-Step Upgrade).

Notes and Constraints

Master nodes can be scaled up only in clusters with two or more Master nodes.
The specifications of the Master node in a BMS cluster cannot be upgraded.
The specifications of local disks cannot be upgraded.
For MRS 1.8.2 or later to a version earlier than MRS 3.x or MRS 3.1.0 or later, see Scaling Up Master Node Specifications (One-Click Upgrade).
For MRS 3.0.5 and a version earlier than MRS 1.8.2, see Scaling Up Master Node Specifications (Step-by-Step Upgrade).
Do not perform other operations on the cluster during the scale-up.
Scale up Master node specifications during off-peak hours to prevent potential service disruptions.

Scaling Up Master Node Specifications (One-Click Upgrade)

Log in to the MRS console.
On the Active Clusters page, select a running target cluster and click its name to go to the cluster details page.
On the Nodes tab page, select Scale Up Specifications in the Operation column of the Master node group. The Scale Up Master Node Specifications page is displayed.
Select the target specifications and click Submit Order. The order has been submitted successfully.

The node specification scale-up takes some time. After the scale-up is successful, the cluster status changes to Running.
- A VM undergoing scale-up is stopped automatically and started again after the scale-up completes.
- Component memory is not automatically upgraded during node specifications scale-up due to differing usage requirements. Adjust the memory configuration manually as required.
Verify the upgrade:

After the upgrade completes, check that the cluster status is Running on the cluster details page. On the Nodes tab, verify that the Master node specifications have been updated to the target specifications.

Scaling Up Master Node Specifications (Step-by-Step Upgrade)

A step-by-step upgrade includes three phases: upgrade preparation, upgrade execution, and post-upgrade verification. This method requires manually stopping and starting services. Perform all steps strictly as directed.

Preparing for Scaling Up Master Node Specifications

Log in to the MRS console.
In the navigation pane on the left, choose Active Clusters, select the cluster for which you want to scale up Master node specifications, and click its name to switch to the cluster details page.
Ensure that the cluster status is Running.
On the Nodes tab page, ensure that all nodes in the cluster are in the Running state.
Log in to Manager and go to the cluster management page. For details, see Accessing MRS Manager.
Choose Cluster > Services > ZooKeeper > Dashboard and ensure that Running Status of the ZooKeeper service is Normal.

Figure 1 ZooKeeper service status
Update service parameter settings as required. For details, see Modifying the Configuration Parameters of an MRS Cluster Component.

Perform this step only once before scaling up the standby Master node.
Choose Cluster > Services > HDFS > Instances.
Record the service IP address of NameNode (Standby). When upgrading the specifications of the active Master node, record the service IP address of NameNode (Active). Figure 2 shows the location of the service IP address.

You need to perform Step 8 to Step 9 to record the IP addresses of the active and standby nodes only when the cluster is an analysis cluster.
Figure 2 Service IP address of the NameNode
On the upper right of the MRS Manager page, check the number displayed next to the icon. If the number is 0, no tasks are running in the cluster.
Click Hosts. If the cluster is an analysis cluster, select the checkbox of the host that matches the NameNode service IP address recorded in Step 9. If the cluster is a streaming cluster, you do not need to distinguish between active and standby nodes. Simply choose hosts for the scale-up.
Choose More > Stop All Instances and wait until all instances are stopped.
- When the node hosting MRS Manager is scaled up, MRS Manager may become temporarily inaccessible due to an active/standby switchover. This is a normal phenomenon. Try logging in to MRS Manager later. If the login attempt fails for a long time, contact O&M personnel for assistance.
- After all roles are stopped, the following alarms may be generated. After the scale-up of Master node specifications is complete and all roles are started, the alarms are automatically cleared.

Scaling Up Master Node Specifications

The following steps scale up the specifications of the standby and active Master nodes in sequence. Do not close or refresh the browser page during the upgrade process.

Log in to the MRS console.
In the navigation pane on the left, choose Active Clusters, select the cluster for which you want to scale up Master node specifications, and click its name to switch to the cluster details page.
On the Nodes tab page, select Scale Up Specifications in the Operation column of the Master node group.
Select the target specifications and click Next.

Make sure the target specification resources are sufficient. If they are not, the active node cannot be scaled up.
On the Confirm page that is displayed, confirm the target node specifications and fees and click OK.
Ensure that all services on the standby Master node have been stopped. (For details, see Step 1 to Step 12 under Preparing for Scaling Up Master Node Specifications.) On the Scale Up Master Node Specifications page, select Are you sure you have stopped all services on the standby Master node? and If not all services are stopped before the scale-up, data saving failure or data damage may occur and click Submit Order.
On the Warning page that is displayed, confirm again that all services on the standby Master node are stopped and click OK to start scaling up the specifications of the standby Master node.

The node specification scale-up takes some time. Please wait. After the scale-up is successful, the cluster status changes to Scaled-up-first. Otherwise, contact O&M personnel.
After the standby Master node has been scaled up successfully, start all services and set parameters on the standby Master node by referring to Step 1 to Step 11 in the Operations After the Master Node Specifications Scale-up part.
After the services on the standby Master node are started, perform an active/standby NameNode switchover. Perform this step only for an analysis cluster and skip this step for a streaming cluster.
1. Access the NameNode web UI of the active and standby nodes separately. For details about how to access the NameNode web UI, see Step 11.
2. In the navigation bar on the NameNode web UI, choose Overview and record the NameNode IDs of the active and standby nodes. Do not close the page after recording.
  Figure 3 NameNode ID of the active node
3. Log in to the ECS of any Master node and run the following command to configure environment variables:
```
source /opt/Bigdata/client/bigdata_env
```
4. If the Kerberos authentication is enabled for the current cluster, run the following command to authenticate the user. If the Kerberos authentication is disabled for the current cluster, skip this step.
```
kinit MRS cluster user
```
  Example:
```
kinit admin
```
5. Run the following command to perform an active/standby NameNode switchover:
```
hdfs haadmin -failover <NameNode ID of the active node> <NameNode ID of the standby node>
```
6. Go to the NameNode web UI page that is not closed in 9.b and refresh the page. You can view that the active/standby NameNode switchover is complete.
  Figure 4 NameNode
Stop all services on the active Master node by referring to Step 1 through Step 12 under Preparing for Scaling Up Master Node Specifications.
On the Scale Up Master Node Specifications page, select I confirm that all services on the standby Master node have been started. and I confirm that all services on the active Master node have been stopped, and click Submit.
On the Confirm page that is displayed, confirm again that all services on the active Master node are stopped and click OK to start scaling up the specifications of the active Master node.

The node specification scale-up takes some time. Please wait. After the scale-up is successful, the cluster status changes to Scaled-up-success. Otherwise, contact O&M personnel.
Start all services and set parameters on the active Master node by referring to Step 1 to Step 11 in the Operations After the Master Node Specifications Scale-up part.
On the Scale Up Master Node Specifications page, select Are you sure you have started all services on the active Master node? and click OK to complete the scale-up.

Operations After the Master Node Specifications Scale-up

The following steps describe how to start all services on the node and verify service status after scaling up node specifications to ensure normal cluster running.

Log in to FusionInsight Manager of the MRS cluster.

For details about how to log in to FusionInsight Manager, see Accessing MRS Manager.
Click Hosts. Check basic information about the host that matches the NameNode service IP address recorded in Step 9 in the Preparing for Scaling Up Master Node Specifications part. If Running Status is Good, and Disk, Memory, and CPU have values, perform Step 9. If any of the preceding conditions are not met, go to the next step.
Log in to the standby Master node remotely. For details, see Logging In to an MRS Cluster Node.
Run the following command to switch to user omm:
```
su - omm
```
Run the following command to start the Agent:
```
sh /opt/Bigdata/nodeagent/bin/start-agent.sh
```
Run the following command to check whether the Agent is started successfully:
```
jps | grep NodeAgent
```
Log in to Manager and go to the cluster management page. For details, see Accessing MRS Manager.
Click Hosts. Check basic information about the host that matches the NameNode service IP address recorded in Step 9 in the Preparing for Scaling Up Master Node Specifications part and ensure that Running Status is Good and Disk, Memory, and CPU have values.

It may take up to 3 minutes for the host status to return to normal after the Agent starts successfully. If the host status remains abnormal for an extended period, contact O&M personnel.
On the MRS Manager page, click Hosts and select the checkbox of the host that matches the NameNode service IP address recorded in Step 9 in the Preparing for Scaling Up Master Node Specifications part.
Choose More > Start All Instances and wait until all instances are started.
Access the NameNode web UI and check the NameNode startup status.
1. On FusionInsight Manager, choose Cluster > Services > HDFS > Dashboard.
2. In the HDFS Summary column, click NameNode of the active or standby node that has been scaled up on the right of NameNode Web UI.
3. Go to the NameNode Web UI page, choose Startup Progress in the navigation bar. After ensuring that Percent Complete is displayed as 100%, go to the next step, as shown in Figure 5.
  Figure 5 NameNode startup status
Perform Step 11 for an analysis cluster and skip this step for a streaming cluster.
Verify the upgrade.

After completing all preceding operations, verify the following items:
- On the MRS console, confirm that the cluster status is Running.
- On the Nodes tab page, verify that the specifications of both the active and standby Master nodes match the target specifications.
- On MRS Manager, confirm that all services display the Normal state and no uncleared alarms exist.

FAQs

What should I do if I cannot log in to MRS Manager during a Master node upgrade?
When the node hosting MRS Manager is being upgraded, an active/standby switchover occurs on MRS Manager. As a result, you cannot log in to MRS Manager for a short period. This is normal behavior. Wait a few minutes and try logging in again. If the login attempt fails for a long time, contact O&M personnel for assistance.
What should I do if an alarm is generated after Master node specification upgrades?
Stopping all roles during the upgrade triggers temporary alarms, such as node fault or service unavailable alarms. These alarms are expected and will automatically clear after the Master node specification upgrade finishes and all roles are started. If an alarm persists after the upgrade completes, check the service status and contact O&M personnel if necessary.
What should I do if component memory configurations are not updated after a one-click upgrade?
A one-click upgrade does not automatically update component memory configurations. You must manually update memory configurations for each component based on your performance requirements. For details, see Modifying the Configuration Parameters of an MRS Cluster Component.

Helpful Links

For details about how to log in to MRS Manager, see Accessing MRS Manager.
For details about how to remotely log in to an MRS cluster node, see Logging In to an MRS Cluster Node.
For details about how to modify cluster component configuration parameters (including memory configuration), see Modifying the Configuration Parameters of an MRS Cluster Component.

Parent Topic: Managing MRS Cluster Nodes

Previous topic: Modifying the Rack Information of an MRS Cluster Node

Next topic: Managing MRS Cluster Node Groups