Alibaba Cloud Elasticsearch supports instance monitoring and allows text message alerting. You can set the alerting thresholds according to your needs.
It is strongly recommended to configure monitoring alerts.
- Cluster status (whether the cluster status indicator is green or red)
- Node disk usage (%) (alerting threshold must be lower than 75%, and cannot exceed 80%)
- Node HeapMemory usage (%) (alerting threshold must be lower than 85%, and cannot exceed 90%)
- Node CPU usage (%) (alerting threshold cannot exceed 95%)
- Node load_1m (reference value: 80% of the number of CPU cores)
- Cluster query QPS (Count/Second) (reference value: practical test result)
- Cluster write QPS (Count/Second) (reference value: practical test result)
Instructions for use
- Elasticsearch console
- CloudMonitor Elasticsearch tab page
Log on to the ES console and go to the ES instance basic information page. Click Cluster Monitor to go to the ES Cloud Monitor module.
Cloud Monitor Elasticsearch tab
Log on to the Alibaba Cloud console using your account, select Cloud Monitor in the product navigator, and choose Elasticsearch from the cloud service monitor menu.
Monitor index configuration
- Choose the area you want to check and click the ES instance ID.
- Create alert policies on the index details page.
On this page, you can check the historical cluster monitoring statistics. The monitoring statistics of the past month are stored. After creating alert policies, you can configure alert monitoring for this instance.
- Enter the policy name and description.
In the following example, the monitoring on disk usage, cluster status, and node HeapMemory usage is configured.
- The cluster status green, yellow, and red match 0.0, 1.0, and 2.0, respectively. Set the values to configure the cluster status alert indexes.
Within the channel silence time, one index can trigger alerting only once.
- Select the alert contact group.
To create a contact group, click Quickly create a contact group.
- Click Confirm to save the alert settings.
Elasticsearch monitoring data is collected five minutes after the instance runs properly. Then the monitoring statistics are displayed.