Prerequisites

• You have obtained the number of OBS cluster nodes in the current environment. For details, see Obtaining the Number of OBS Cluster Nodes in the Current Environment.
• You have logged in to ManageOne Maintenance Portal.

Procedure

1. Check the number of OBS cluster nodes allowed for the ManageOne VM management scale in the current environment by referring to Table 3-235.
   Table 3-235 Number of OBS cluster nodes allowed for ManageOne management scales

   ManageOne Deployment Mode	VM Management Scale	OBS Cluster Nodes	Number of Slots for OBS Log Analysis	Number of Available Slots
   15 nodes	500~2000	Not supported	Not supported	Not supported
   	2000~5000	6	14	26
   	5000~10000	9	15	26
   	10000~30000	12	25	58
   	30000~100000	15	30	55
   12 nodes	500~2000	Not supported	Not supported	Not supported
   	2000~5000	3	14	10
   	5000~10000	6	14	10
   	10000~30000	8	15	10

2. Check whether the number of OBS cluster nodes in the current environment is greater than the number of OBS cluster nodes obtained in Step 1 in the current environment.
   • If yes, obtain the minimum management scale corresponding to the actual number of OBS cluster nodes by referring to Table 3-235, expand the ManageOne VM management scale in the current environment to this scale based on HUAWEI CLOUD Stack 8.1.1 Capacity Expansion Guide, and go to Step 3.
   • If no, go to Step 3.

3. If ManageOne is deployed on 12 nodes, add Flink compute resources by following Adding Flink Compute Resources.

   If ManageOne is deployed on 15 nodes, you do not need to add Flink compute resources. Skip this step and perform Step 4 and Step 5.

4. Enable the log metric monitoring function.
   1. Choose System > System Integration from the main menu.
   2. In the navigation pane, choose Data Channel.
   3. Enter LogMetricConfig in the search box in the upper right corner of the page and click .
   4. Click the LogMetricConfig group name. The View Configuration dialog box is displayed.
   5. Check whether the value of isSupportLogMetric is true.
      • If yes, the log metric monitoring function has been enabled. Go to Step 5.
      • If no, click Cancel to close the View Configuration dialog box.
   6. Click Modify in the Operation column of the LogMetricConfig group.
   7. In the displayed dialog box, click OK.
   8. In the Modify Configuration dialog box, change the value of isSupportLogMetric to true.
   9. Click OK.

5. View monitoring data of OBS log metrics.

   After the log metric monitoring function is enabled, log metric data is not reported immediately. You need to wait for about 30 minutes.

   1. Choose Monitor > Resource Monitoring from the main menu.
   2. Choose System Resources in the navigation pane and click the Cloud Service tab to switch to the cloud service monitoring list.
   3. Enter OBS in the search box above the cloud service monitoring list.
   4. After the search is successful, click the OBS cloud service name. The OBS cloud service overview page is displayed.
   5. In the navigation pane, choose Basic Information > Monitoring.
   6. Select the OBS root node in the service tree and view the Business Metric page on the right.
   7. Check whether values are displayed in the chart below.
      • If yes, the OBS log metric monitoring function is enabled successfully.
      • If no, contact technical support for assistance.

Obtaining the Number of OBS Cluster Nodes in the Current Environment

1. Choose Resource Topology > Resource Management from the main menu.
2. In the navigation pane, choose Resource List.
3. Click the System Resources tab on the All Resources or Resource Pools tab page.
4. Select Cloud Service > Microservice Instance for Resources.
5. In the resource list, set Name to osc_ to search for resources.
6. Calculate the number of search records whose names start with osc_, that is, the number of OBS cluster nodes in the current environment.

Checking the Total and Used Physical Memory of MODataPipelineService Nodes

1. Use a browser to log in to ManageOne Deployment Portal.

   URL: https://{Floating IP address of ManageOne Deployment Portal}:31945, for example, https://192.168.0.1:31945

   Default username: admin

2. Choose Maintenance > Operation and Maintenance Management > Panoramic Monitoring from the main menu of ManageOne Deployment Portal.
3. In the navigation pane, choose Service Monitoring.
4. Click in the upper left corner of the page and choose ManageOne.
5. Click the Processes tab.
6. Enter MODataPipelineService in the search box in the upper left corner and press Enter.
7. In the search result, view the node names of all MODataPipelineService processes.
8. In the navigation pane, choose Node Monitoring.
9. In the search box in the upper left corner, enter the node name queried in Step 7 and press Enter.
10. In the search result, view the total and used physical memory of each node.

Adding Flink Compute Resources

The OBS log analysis function consumes a large number of Flink compute resources. Therefore, before enabling the log metric monitoring function, you need to scale out Flink clusters in the current environment.

1. Check whether the number of slots to be added in the current environment is greater than the maximum number of slots that can be added.
   • The number of slots to be added in the current environment is a value in the Number of Slots to Be Added for OBS Log Analysis column in Table 3-236.
   • For 12 nodes, the maximum number of slots that can be added is twice a value in the Maximum Number of Slots That Can Be Added for a Single Node column in Table 3-236.
   • If yes, expand the ManageOne VM management scale in the current environment to the next scale by referring to Table 3-236, and then go to Step 2.

     If the management scale of ManageOne VMs before scale-out is 10,000 to 30,000 in 12-node mode, the next scale is 30,000 to 100,000 in 15-node mode.

   • If no, go to Step 2.

2. Calculate the total memory after slots are added (Used physical memory + Memory to be added).
   Check whether the calculated total memory exceeds 80% of the total VM memory.

   • Obtain the total VM memory and used physical memory from Checking the Total and Used Physical Memory of MODataPipelineService Nodes.
   • Memory to be added indicates a value in the Memory to Be Added for OBS Log Analysis column in Table 3-236.

   • If yes, expand the ManageOne VM management scale in the current environment to the next scale by referring to Table 3-236, and then go to Step 3.

     If the management scale of ManageOne VMs before scale-out is 10,000 to 30,000 in 12-node mode, the next scale is 30,000 to 100,000 in 15-node mode.

   • If no, go to Step 3.

3. Modify the configuration file of the node where MODataPipelineService is deployed.
   1. Use PuTTY to log in to a node where MODataPipelineService is deployed, as the sopuser user over SSH.
   2. Run the following command to switch to the root user:

      sudo su root

   3. Run the following command to go to the directory where the target file is stored:

      cd /opt/oss/Product/apps/MODataPipelineService/flink-1.11.3/conf

   4. Run the vi command to open the flink-conf.yaml file and modify the following items:
      • taskmanager.numberOfTaskSlots: Change the value to the total number of slots after slots are added.
      • taskmanager.memory.task.heap.size: Change the value to the total memory calculated in Step 2.
   5. After modification, press Esc to exit the editing mode and enter :wq! to save the settings.

4. Restart MODataPipelineService on the current node.

   sudo -uossadm /opt/oss/manager/agent/bin/ipmc_adm -cmd restartapp -app MODataPipelineService

5. Repeat Step 3 and Step 4 on other nodes where MODataPipelineService is deployed.