view the host monitoring metrics of application instances in a Kubernetes cluster - Enterprise Distributed Application Service

When applications experience slow responses or unexpected failures, you need visibility into the underlying host resources to determine whether the issue is infrastructure-related. Enterprise Distributed Application Service (EDAS) collects 19 host-level metrics across CPU, memory, disk, load, network traffic, and network packets, so you can identify resource bottlenecks and troubleshoot performance issues without leaving the console.

Monitored metrics

EDAS tracks metrics across six categories. Each metric appears as a time series curve on the HOST Monitoring tab.

CPU

Metric	Description
`total`	Overall CPU utilization across all cores.
`cpu_sys`	Percentage of CPU time spent on kernel (system) processes. A sustained high value may indicate excessive system calls or I/O operations.
`cpu_user`	Percentage of CPU time spent on user-space processes, including your application code.
`cpu_iowait`	Percentage of CPU time waiting for I/O operations to complete. A high value typically points to disk or network bottlenecks.

Memory

Metric	Description
`total`	Total physical memory on the host.
`mem_used`	Memory actively in use by processes. Track this metric to detect memory pressure.
`mem_free`	Unused memory not allocated to any process.
`mem_cach`	Memory used by the OS page cache to speed up disk reads. This memory is reclaimable and released when applications need it.
`mem_buff`	Memory used by kernel buffers for I/O operations. Like page cache, this memory is reclaimable.

A low mem_free value does not always indicate a problem. Linux aggressively uses available memory for mem_cach and mem_buff to improve performance. Evaluate mem_used alongside mem_free to determine actual memory pressure.

Disk

Metric	Description
`total`	Total system disk capacity.
`SystemDiskUsed`	Disk space currently in use.
`SystemDiskFree`	Available disk space. Monitor this metric to prevent disk-full errors that can cause application crashes or failed log writes.

Load

Metric	Description
`systemLoad`	System load average. If this value consistently exceeds the number of CPU cores on the host, the host may be overloaded.

Network traffic

Metric	Description
`traffic_bytin`	Inbound network traffic in bytes.
`traffic_bytout`	Outbound network traffic in bytes.

Network packets

Metric	Description
`traffic_pktin`	Number of inbound packets received.
`traffic_pktout`	Number of outbound packets sent.
`traffic_pkterr`	Number of error packets. A non-zero value may indicate network interface issues, driver problems, or packet corruption.
`traffic_pktdrp`	Number of dropped packets. Drops can result from buffer overflow, congestion, or firewall rules. Investigate if this value increases steadily.

View host monitoring metrics

Prerequisites

Before you begin, make sure that you have:

An application deployed on a Kubernetes Cluster in EDAS
Access to the EDAS console

Procedure

Log on to the EDAS console.
In the left-side navigation pane, choose Application Management > Applications.
In the top navigation bar, select a region. In the upper part of the page, select a microservice namespace.
From the Cluster Type drop-down list, select Kubernetes Cluster.
Find your application and click its name in the Application Name column.
In the left-side navigation pane of the application details page, choose Monitor > Instance Details.
On the Instance Details page, select the node to inspect and click the HOST Monitoring tab.

Interact with charts

Each metric category (CPU, Memory, Disk, Load, Network Traffic (Bytes), and Network Packets (Count)) appears as a separate time series chart.

Show or hide a metric: Click the metric name (for example, cpu_sys) to toggle its visibility.
Minimum visibility: Each chart requires at least one visible metric. If only one metric is displayed, you cannot hide it.

Troubleshoot common issues

Use the following table to identify and respond to common resource problems:

Symptom	Likely cause	Recommended action
High `cpu_sys` with high `cpu_iowait`	Disk or network I/O bottleneck	Check disk throughput and network latency. Consider upgrading to a higher I/O instance specification.
Low `mem_free` with high `mem_used`	Memory pressure or memory leak	Scale up the instance specification or add more application instances. Review your application for memory leaks.
Low `SystemDiskFree`	Disk running out of space	Clean up unnecessary files such as old logs. Expand disk capacity or mount additional storage.
`systemLoad` exceeds CPU core count	CPU overload	Add more application instances to distribute the load, or upgrade to an instance type with more CPU cores.
Increasing `traffic_pkterr` or `traffic_pktdrp`	Network interface or configuration issues	Check the network configuration of the host and the associated Kubernetes node. Verify firewall rules and network driver status.

For application-level performance analysis, use the application monitoring features under Monitor in the EDAS console.