Use health diagnostics, system events, metrics, and alerts to detect and resolve ECS instance issues before they affect your business.
Handle system events promptly
The system sends system events when it performs O&M actions or detects issues that might affect instance operations. Each event includes recommended actions and a timeline. Handle system events promptly to prevent issues such as instance restarts or stops from affecting your business. See Overview of ECS system events.
The following figures show a system event example for an expiring subscription instance.
In Message Center, enable internal messages for ECS expiration, product O&M, and ECS failure notifications. Otherwise, you will not receive system events in the ECS console. The following figure shows the Settings page.
View instance metrics
Alibaba Cloud collects and displays real-time and historical instance metrics. Use these metrics to determine whether the instance runs normally. For example, consistently high CPU utilization might indicate an abnormal process or an undersized instance.
View instance metrics on the Monitoring tab of the instance details page in the ECS console, or on the Host Monitoring page in the CloudMonitor console. See View instance monitoring information and Overview of host monitoring.
-
Metrics on the instance details page in the ECS console:
-
Compute, storage, and network resource usage, such as CPU utilization, disk read/write status, and packet counts.
-
CPU credit usage for burstable instances.
-
-
Metrics on the OS Monitoring tab of the Host Monitoring page in the CloudMonitor console:
-
Compute, storage, and network resource usage, such as CPU utilization, disk read/write status, and packet counts.
-
Active process information within the instance.
-
GPU memory usage for GPU-accelerated instances.
-
Set up alert notifications
Set CloudMonitor alert rules for specific events or instance metrics. When an event occurs or a metric becomes abnormal, CloudMonitor notifies contacts by email or other methods. See Subscribe to ECS system event notifications and Set alert rules for an ECS instance.
The following figure shows an alert rule for an event.
The following figure shows an alert rule for an instance metric.
