Amazon CloudWatch RDS monitoring can help you identify and avoid capacity or other issues with the metadata store for the cluster. With the Amazon CloudWatch integration, enabling metadata store monitoring for your cluster allows Imply support to track the health of your clusters and get notifications of potential issues. Specifically, alerts are triggered for these conditions:
- CPU usage > 80% of capacity
- Memory usage > 80% of capacity
- Storage usage > 80% of capacity
- Connections > 80% of max_connections
Note that using Amazon CloudWatch with Imply may entail additional costs, which are not included in the Imply subscription. See CloudWatch pricing for details.
Enabling Metadata monitoring
Metadata monitoring is enabled by default for new clusters created after the availability of Imply version 3.4 (July, 2020).
For clusters that existed prior to that time, metadata store monitoring is disabled by default to avoid the possibility of interfering with existing settings or validated configurations.
You can enable metadata store monitoring as follows.
- As the Imply role in the AWS account needs permission to interact with Amazon CloudWatch, in the IAM settings, create a policy for Imply that enables the following permissions for the CloudWatch service:
- Delete alarms (cloudwatch:DeleteAlarms): Required to delete an alarm.
- DescribeAlarms (cloudwatch:DescribeAlarms): Required to retrieve alarm information by name.
- PutMetricAlarm (cloudwatch:PutMetricAlarm): Required to create or update an alarm.
- Enable metadata store monitoring as follows:
- Navigate to the Cluster Page > Setup > Advanced Config.
- In the Metadata Storage section, set the metadata store monitoring value to Enabled.
- Click Apply changes.