ECS Event Handling Suggestions

ECS Event Overview

On the CCE console, you can use ECSs as nodes to create highly available Kubernetes clusters. During routine O&M, Huawei Cloud proactively forecasts and prevents hardware and software faults on hosts accommodating ECSs.

If the faults cannot be avoided, the system generates and reports ECS events (like ECS redeployment and system maintenance) for the affected ECSs to minimize the impacts of resource unavailability or performance deterioration.

You can manage ECS events on the ECS console. For details, see ECS Event Overview .

Scenarios

During ECS event handling, the ECS may be unavailable, which makes the node and pods cannot run properly. To prevent any disruption to your services, when the system generates ECS events for your node, you are advised to migrate services running on the node to other available nodes in advance before handling the ECS events.

Procedure

1. Perform pre-processing.

   Before handling ECS events, you need to migrate services to other available nodes and then isolate the faulty node.

   You can refer to Draining a Node to migrate pods running on the node and then isolate the node. Alternatively, you can manually migrate pods running on the node and then refer to Managing Node Taints to add a NoSchedule taint to the node for its isolation.

   You can also migrate services to other available nodes, create a node, and delete the faulty node to skip subsequent steps.

2. Handle events.

   Go to the ECS console to query events and handle ECS events by event type.

3. Perform post-processing.

   After processing the ECS events, cancel the node isolation.

   To restore the node to the schedulable state, you can click More > Enable Schedulingon the Nodes page or remove the NoSchedule taint of the node by referring to Managing Node Taints.