The Kubernetes container monitoring feature in Cloud Monitor is being phased out. Migrate to Managed Service for Prometheus, which provides equivalent monitoring coverage.
Use Cloud Monitor to track CPU, memory, and network usage across your ACK clusters. Cloud Monitor automatically collects metrics for all ACK clusters in your Alibaba Cloud account, giving you a centralized view across multiple regions.
Prerequisites
Before you begin, make sure that:
-
The metrics-server component in your cluster is V0.3.8.5 or later
-
ACK managed cluster: Manage components
-
ACK dedicated cluster running Kubernetes 1.12 or earlier: Update the metrics-server component before you update the Kubernetes version to 1.12
-
-
If metrics-server cannot be updated to V0.3.8.5 or later, use the previous version of resource monitoring instead
How monitoring works
Cloud Monitor provides Kubernetes-native monitoring at three layers:
| Layer | What is monitored |
|---|---|
| Host infrastructure | Node-level CPU and memory utilization |
| Container (PaaS) | Pod CPU and memory usage; network resources |
| Kubernetes scheduling | Working memory that affects pod scheduling decisions (dedicated to container working memory, helping distinguish container memory usage from host memory usage) |
At the cluster level, Cloud Monitor surfaces alerts, node counts, and pod and node resource usage across namespaces, nodes, workloads, and pods.
Enable Cloud Monitor for an ACK cluster
See Enable the Kubernetes container monitoring feature for an ACK cluster.
View monitoring data
-
Log on to the Cloud Monitor console.
-
In the left-side navigation pane, choose Cloud Service Monitoring > Container Service ACK.
-
On the Container Service Monitoring page, find the cluster and click its name or click View Details in the Actions column.
If this is the first time you access Cloud Monitor for this cluster, click Authorize to complete authorization before the cluster details page opens.
-
On the cluster details page, review monitoring data in the Cluster overview, Node, Namespace, Workload, and Alert Rules sections. For details about each section, see View monitoring data.
Set up alert rules
Step 1: Create an alert contact and alert contact group
-
Log on to the Cloud Monitor console.
-
In the left-side navigation pane, choose Alerts > Alert Contacts.
-
Create an alert contact and add it to an alert contact group. See Create an alert contact or alert contact group.
Step 2: Create an alert rule
-
Log on to the Cloud Monitor console.
-
In the left-side navigation pane, choose Container Service Monitoring > Container Service ACK.
-
On the Container Service Monitoring page, find the cluster and click View Alert Rules in the Actions column.
-
Click Create Alert Rule.
-
In the Create Alert Rule panel, configure the following parameters:
Parameter Description Resource Range The scope to which the alert rule applies. Valid values: Cluster, Node, or Container Group (pod). For Container Group (pod), select the namespace and then choose a workload type (Deployment, StatefulSet, DaemonSet, Job, or CronJob) and one or more pods. Rule Description The condition that triggers the alert. Set the metric, threshold, and alert level. For available pod metrics, see ACK (new version). Mute Period The interval before Cloud Monitor resends a notification for the same alert. Valid values: 5 Minutes, 15 Minutes, 30 Minutes, 60 Minutes, 3 Hours, 6 Hours, 12 Hours, 24 Hours. Cloud Monitor does not resend a notification if the alert level is unchanged within the mute period. Cloud Monitor resends a notification if the alert level changes (for example, the alert is cleared) within the mute period, or if the metric value reaches the threshold again after the mute period ends. Cloud Monitor does not send alert notifications when the resource is in the Mute state (single resource in mute period) or the Notification Disabled state (multiple resources in mute period). Effective Period The time window during which the alert rule is active. Cloud Monitor only monitors the resource within this period. Alert Callback An Internet-accessible URL to which Cloud Monitor sends HTTP POST requests when an alert fires. Only the HTTP protocol is supported. See Use the alert callback feature to send notifications about threshold-triggered alerts. Alert Contact Group The contact groups that receive alert notifications. Each group can contain one or more alert contacts. See Create an alert contact or alert contact group. -
Click OK. The new rule appears in the Alert Rules section. To manage existing rules, see Manage alert rules.
Verify
-
In the left-side navigation pane, choose Alerts > Alert History.
-
On the Alert History page, check the alert trend and detailed alert records.
Alert scenarios
The following table describes common alerting scenarios and the corresponding Resource Range and configuration required for each.
| Scenario | Description | Configuration |
|---|---|---|
| Cluster or node resource usage exceeds threshold | Alert when the overall cluster or any node crosses the resource limit to prevent service disruption | Set Resource Range to Cluster or Node. For Node, select all nodes so the rule fires when any node exceeds the threshold. |
| Any pod in the cluster exceeds threshold | Alert when any pod causes resource pressure, making it easier to identify the offending workload | Set Resource Range to Container Group (pod). Select All from both the Namespace and Container Group (pod) lists. |
| Any pod in a specific namespace exceeds threshold | Alert on a per-namespace basis in multi-tenant clusters where applications share the same cluster | Set Resource Range to Container Group (pod). Select the target namespace, then select All from the Container Group (pod) list. |
| Any pod of a specific workload exceeds threshold | Alert on a per-workload basis, for example, when a Deployment's resource usage spikes | Set Resource Range to Container Group (pod). Select the namespace and the target workload (Deployment, StatefulSet, DaemonSet, Job, or CronJob), then select All from the Container Group (pod) list. |
Use the previous version of resource monitoring
If metrics-server cannot be updated to V0.3.8.5 or later, use the following steps to access resource monitoring in the ACK console.
-
Log on to the ACK console. In the left-side navigation pane, click Clusters.
-
On the Clusters page, find the cluster and click its name. In the left-side pane, choose Workloads > Deployments.
-
On the Deployments page, find the Deployment and click Monitor in the Actions column.
-
Review monitoring data on the Deployment Application, Container group list, and Container group hotspot tabs.
-
(Optional) In the left-side navigation pane, choose Alerts > Alert Rules to configure alert rules.
Group-based metric names start with
group; instance-based metric names start withpod.
FAQ
No monitoring data appears for my ACK cluster in Cloud Monitor
See What do I do if no data exists in an ACK cluster? for troubleshooting steps.
What's next
-
Overview — learn about the full scope of Cloud Monitor's Kubernetes monitoring capabilities
-
View monitoring data — explore each monitoring section in detail
-
Managed Service for Prometheus — migrate to the recommended monitoring solution