Monitoring Clusters Using Cloud Eye

Function

This section describes how to check cluster metrics on Cloud Eye. By monitoring cluster running metrics, you can identify the time when the database cluster is abnormal and analyze potential activity problems based on the database logs, improving database performance. This section describes the metrics that can be monitored by Cloud Eye as well as their namespaces and dimensions. You can use the management console or APIs provided by Cloud Eye to query the monitoring metrics and alarms generated by GaussDB(DWS).

Namespace

SYS.DWS

Cluster Monitoring Metrics

With the GaussDB(DWS) monitoring metrics provided by Cloud Eye, you can obtain information about the cluster running status and performance. This information will provide a better understanding of the node-level information.

Table 1 describes GaussDB(DWS) monitoring metrics.

**Table 1** GaussDB(DWS) monitoring metrics
Metric ID	Name	Description	Value Range	Monitored Object	Monitoring Period (Raw Data)
dws001_shared_buffer_hit_ratio	Cache Hit Ratio	Ratio of requested data that already exists in the cache. It the ratio of the amount of data that already exists in the cache to the total amount of requested data. A higher cache hit ratio means higher cache usage of the system, fewer times that data needs to be read from the disk or network, and faster system response speed. Unit: Percent	0% to 100%	Data warehouse cluster	4 minutes
dws002_in_memory_sort_ratio	In-memory Sort Ratio	Ratio of the extra memory space used by the sorting algorithm to the memory space occupied by the sorted data. In a merge sort, for example, the size of the merge buffer is often proportional to the size of the sorted data, so the in-memory ratio is usually between 10% and 50%. Unit: Percent	0% to 100%	Data warehouse cluster	4 minutes
dws003_physical_reads	File Reads	Total number of database file reads	> 0	Data warehouse cluster	4 minutes
dws004_physical_writes	File Writes	Total number of database file writes	> 0	Data warehouse cluster	4 minutes
dws005_physical_reads_per_second	File Reads per Second	Number of database file reads per second	≥ 0	Data warehouse cluster	4 minutes
dws006_physical_writes_per_second	File Writes per Second	Number of database file writes per second	≥ 0	Data warehouse cluster	4 minutes
dws007_db_size	Data Volume	Total size of data in the database, in MB	≥ 0 MB	Data warehouse cluster	4 minutes
dws008_active_sql_count	Active SQL Count	Number of active SQLs in the database	≥ 0	Data warehouse cluster	4 minutes
dws009_session_count	Session Count	Number of sessions that access the database	≥ 0	Data warehouse cluster	4 minutes
dws010_cpu_usage	CPU Usage	CPU usage of each node in a cluster, in percentage	0% to 100%	Data warehouse node	1 minute
dws011_mem_usage	Memory Usage	Memory usage of each node in a cluster, in percentage NOTE: After the console is upgraded to 8.3.0.202, the memory usage includes the memory occupied by the cache. Therefore, the value of this metric increases compared with that before the upgrade.	0% to 100%	Data warehouse node	1 minute
dws012_iops	IOPS	Number of I/O requests processed by each node in the cluster per second	≥ 0	Data warehouse node	1 minute
dws013_bytes_in	Network Input Throughput	Data input to each node in the cluster per second over the network Unit: byte/s	≥ 0 bytes/s	Data warehouse node	1 minute
dws014_bytes_out	Network Output Throughput	Data sent to the network per second from each node in the cluster Unit: byte/s	≥ 0 bytes/s	Data warehouse node	1 minute
dws015_disk_usage	Disk Usage	Disk usage of each node in a cluster, in percentage	0% to 100%	Data warehouse node	1 minute
dws016_disk_total_size	Total Disk Size	Total disk space of each node in the cluster Unit: GB	100 to 2000 GB	Data warehouse node	1 minute
dws017_disk_used_size	Used Disk Space	Used disk space of each node in the cluster Unit: GB	0 to 3600 GB	Data warehouse node	1 minute
dws018_disk_read_throughput	Disk Read Throughput	Data volume read from each disk in the cluster per second Unit: byte/s	≥ 0 bytes/s	Data warehouse node	1 minute
dws019_disk_write_throughput	Disk Write Throughput	Data volume written to each disk in the cluster per second Unit: byte/s	≥ 0 bytes/s	Data warehouse node	1 minute
dws020_avg_disk_sec_per_read	Average Time per Disk Read	Average time used each time when a disk reads data Unit: second	> 0s	Data warehouse node	1 minute
dws021_avg_disk_sec_per_write	Average Time per Disk Write	Average time used each time when data is written to a disk Unit: second	> 0s	Data warehouse node	1 minute
dws022_avg_disk_queue_length	Average Disk Queue Length	Average I/O queue length of a disk	≥ 0	Data warehouse node	1 minute
dws_024_dn_diskio_util	DN I/O usage	Average disk I/O usage of DNs in a cluster	0% to 100%	Data warehouse instance	1 minute

Dimensions

Key	Value
datastore_id	Data warehouse cluster ID
dws_instance_id	Data warehouse node ID

Cluster and Node Monitoring Information

Log in to the GaussDB(DWS) console and choose Clusters > Dedicated Clusters.
View the cluster information. In the cluster list, click View Metric in the Operation column where a specific cluster resides. The Cloud Eye management console is displayed. By default, the cluster monitoring information on the Cloud Eye management console is displayed.

Additionally, you can specify a specific monitoring metric and the time range to view the performance curve.
View the node information. Click to return to the Cloud Eye management console. On the Data Warehouse Nodes tab page in the right pane, you can view metrics of each node in the cluster.

Additionally, you can specify a specific monitoring metric and the time range to view the performance curve.

Cloud Eye also supports the ability to compare the monitoring metrics of multiple nodes. For details, see Comparing the Monitoring Metrics of Multiple Nodes.

Comparing the Monitoring Metrics of Multiple Nodes

In the navigation pane of the Cloud Eye management console, choose Dashboards > My Dashboards. Click the name of the dashboard for which you want to add a graph. On the My Dashboards page that is displayed, click Add Graph.

On the Add Graph page, you can select Line Chart or Bar Chart to display the graph. After confirming that the information is correct, click OK.

For example, select Line Chart and One View for Multiple Metrics to compare the CPU usage of three GaussDB(DWS) nodes. The following table describes the parameters.

Click to enlarge

**Table 2** Configuration example
Parameter	Example Value
Resource Type	DWS
Dimension	Data Warehouse Node
Monitored Object	dws-demo-dws-cn-cn-2-1 dws-demo-dws-cn-cn-1-1 dws-demo-dws-dn-1-1
Metric	CPU Usage

Click OK.

On the selected My Dashboards page, you can view the metric trend on the newly added monitoring graph. You can click the zoom in button to zoom in and view detailed metric comparison data.

Creating Alarm Rules

GaussDB(DWS) enables you to customize alarm rules for monitoring specific objects and notification policies, ensuring you stay informed about its running status in a timely manner.

A GaussDB(DWS) alarm rule includes the alarm rule name, monitored object, metric, threshold, monitoring interval, and whether to send a notification. This section describes how to set GaussDB(DWS) alarm rules.

Log in to the GaussDB(DWS) console.
In the navigation pane, choose Clusters > Dedicated Clusters.
Locate the row containing the target cluster, click View Metric in the Operation column to enter the Cloud Eye management console and view the GaussDB(DWS) monitoring information.

The status of the target cluster must be Available. Otherwise, you cannot create alarm rules.
In the left navigation pane of the Cloud Eye management console, choose Alarm Management > Alarm Rules.
On the Alarm Rules page, click Create Alarm Rule in the upper right corner.

On the Create Alarm Rule page, set parameters as prompted.

Configure the rule name and description.

Configure the alarm parameters as prompted.

**Table 3** Configuring alarm parameters
Parameter	Description	Example Value
Resource Type	Name of the cloud service resource for which the alarm rule is configured.	Data Warehouse Service
Dimension	Metric dimension of the alarm rule. You can select Data Warehouse Nodes or Data Warehouses.	Data Warehouse Node
Monitoring Scope	Resource scope to which an alarm rule applies. Select Specific resources and select one or more monitoring objects. For GaussDB(DWS), select the cluster ID or node ID in the dialog box that is displayed.	Specific resources
Trigger Rule	You can select an associated template, use an existing template or create a custom template as required.	Create manually
Template	This parameter is valid only when Use template is selected. Select the template to be imported. If no alarm template is available, click Create Custom Template to create one that meets your requirements.	-
Alarm Policy	This parameter is valid only when Create manually is selected. Set the policy that triggers an alarm. For example, trigger an alarm if the CPU usage equals to or is greater than 80% for 3 consecutive periods. Table 1 lists the GaussDB(DWS) monitoring metrics.	-
Alarm Severity	Severity of an alarm. Valid values are Critical, Major, Minor, and Informational.	Major

Configure the alarm notification parameters as prompted.

**Table 4** Configuring alarm notifications
Parameter	Description	Example Value
Alarm Notification	Whether to notify users when alarms are triggered. Notifications can be sent as emails or text messages, or HTTP/HTTPS requests sent to the servers. You can enable (recommended) or disable Alarm Notification.	Enable
Validity Period	Cloud Eye sends notifications only within the validity period specified in the alarm rule. For example, if Validity Period is set to 00:00-8:00, Cloud Eye sends notifications only within 00:00-8:00.	-
Notification Object	Name of the topic to which the alarm notification is sent. If you enable Alarm Notification, you need to select a topic. If no desired topics are available, create one first, whereupon the SMN service is invoked. For details about how to create a topic, see the Simple Message Notification User Guide. For details about how to create a topic, see the Simple Message Notification User Guide.	-
Trigger Condition	Condition for triggering the alarm. You can select Generated alarm, Cleared alarm, or both.	-

After the configuration is complete, click Next.
After the alarm rule is created, if the metric data reaches the specified threshold, Cloud Eye will immediately inform you that an exception has occurred.