VM Monitoring Metric Description

JD Cloud provide instances with the following monitoring indicators which can be divided into three categories according to the collected and reported preconditions:

• They are collected by the host machine where the instance is located, rather than the monitoring plug-in in Virtual Machines. There are 4 indicators of this type, whose Chinese or English display names have suffix ‘(Host)’, including:
  • Disk Read Throughput (Host): vm.disk.bytes.read
  • Disk Write Throughput (Host): vm.disk.bytes.write
  • Network Input Bandwidth (Host): vm.network.bytes.incoming
  • Network Output Bandwidth (Host): vm.network.bytes.outgoing
• They are collected by the public system components in Virtual Machines. All previous version components support collection and get data as long as these components are not uninstalled. There are 2 indicators of this type, including:
  • CPU Util: cpu_util
  • Memory usage: memory.usage
• The remaining indicators are collected by the public system components in Virtual Machines. Only JCS-Agent components of which versions are not lower than '3.0.989' support collection. If you can view this type of indicators on the monitoring page, it means that the version of system components in your current environment is too low, please refer to Monitoring Plug-in Installation Description at the bottom of this context for installation.

Indicator Details

The servicecode of product line monitoring data: vm. The indicator details are as follows:

Instance

metric	Chinese Name	English Name	Unit	Description
cpu_util	CPU Usage	CPU Usage	%
memory.usage	Memory Usage	Memory Usage	%
vm.memory.used.bytes	Memory Used	Memory Used	Bytes
vm.avg.load1	CPU Average Load 1min	CPU Average Load 1min	-
vm.avg.load5	CPU Average Load 5min	CPU Average Load 5min	-
vm.avg.load15	CPU Average Load 15min	CPU Average Load 15min	-
vm.disk.dev.bytes.read	Disk Read Throughout	Disk Read Throughout	Bps
vm.disk.dev.bytes.write	Disk Write Throughout	Disk Write Throughout	Bps
vm.disk.dev.io.read	Disk Read IOPS	Disk Read Requests	Count/s
vm.disk.dev.io.write	Disk Write IOPS	Disk Write Requests	Count/s
vm.network.dev.bytes.in	Network Inbound Rate	Network Inbound Rate	bps
vm.network.dev.bytes.out	Network Outbound Rate	Network Outbound Rate	bps
vm.network.dev.packets.in	Network Inbound Packets	Network Inbound Packets	pps
vm.network.dev.packets.out	Network Outbound Packets	Network Outbound Packets	pps
vm.netstat.tcp.established	TCP Connections	TCP Connections	Count	Bare metal has no such metric
vm.disk.bytes.read	Disk Read Throughout(host)	Disk Read Throughout(host)	Bps	Bare metal has no such metric
vm.disk.bytes.write	Disk Write Throughout(host)	Disk Write Throughout(host)	Bps	Bare metal has no such metric
vm.network.bytes.incoming	Network Inbound Rate(host)	Network Inbound Rate(host)	bps	Bare metal has no such metric
vm.network.bytes.outgoing	Network Outbound Rate(host)	Network Outbound Rate(host)	bps	Bare metal has no such metric

Device - Disk

Amount

Linux reports as per mount points;Windows reports as per disk symbol.tag key is mountPoint, and value is‘/’,‘C: ’...

metric	Chinese Name	English Name	Unit	Description
vm.disk.dev.used	Disk Usage	Disk Usage	%
vm.disk.dev.used.bytes	Disk Used	Disk Used	Bytes
vm.disk.dev.inode.used	Disk inode Usage	Disk inode Usage	%

Read-write

Linux reports as per partition device name;Windows reports as per disk symbol. Tag key is devName, and value is‘/dev/vda1’,‘C: ’...

metric	Chinese Name	English Name	Unit	Description
vm.disk.dev.bytes.read	Disk Read Throughout	Disk Read Throughout	Bps
vm.disk.dev.bytes.write	Disk Write Throughout	Disk Write Throughout	Bps
vm.disk.dev.io.read	Disk Read IOPS	Disk Read Requests	Count/s
vm.disk.dev.io.write	Disk Write IOPS	Disk Write Requests	Count/s

Device - Network

key of tag is devName. Its value is 'eth0','eth1'...'eth7

metric	Chinese Name	English Name	Unit	Description
vm.network.dev.bytes.in	Network Inbound Rate	Network Inbound Rate	bps
vm.network.dev.bytes.out	Network Outbound Rate	Network Outbound Rate	bps
vm.network.dev.packets.in	Network Inbound Packets	Network Inbound Packets	pps
vm.network.dev.packets.out	Network Outbound Packets	Network Outbound Packets	pps

Device - GPU

Only Machine with GPU instance type has such metric. tag keyisgpu_index, and value is'0','1','2' and‘3’

metric	Chinese Name	English Name	Unit	Description
vm.gpu.util.gpu	GPU Core Usage	GPU Core Usage	%
vm.gpu.util.mem	GPUMemory Usage	GPU Memory Usage	%
vm.gpu.used.mem.bytes	GPU Memory Used	GPU Memory Used	Bytes
vm.gpu.power	GPU Power Used	GPU Power	w
vm.gpu.temperature	GPU Temperature	GPU Temperature	℃
vm.gpu.util.encoder	GPU Encoder Usage	GPU Encoder Usage	%
vm.gpu.util.decoder	GPU Decoder Usage	GPU Decoder Usage	%

Monitoring Plug-in Installation Description

The collection and report of VM Monitoring data rely on 'MonitorPlugin' Plug-in in the public image system component 'JCS-Agent'. The public image which was upgraded from May to July 2019 has installed by default the upgrade tool 'ifrit' to implement the automatic upgrade of JCS-Agent.
If your current instance didn't install JCS-Agent or installed a lower version that has no automatic upgrade ability, you can directly install ifrit provided that you can ensure the early-stage system components cloud-init and QGA were uninstalled. Within 10 minutes after the installation, JCS-Agent will be automatically installed / updated to the latest version.

• For the uninstallation method of cloud-init and QGA and the installation method of Ifrit, refer to: Public Image System Component-JCS-Agent
• JCS-Agent Version View Method:
  • Linux： ps -ef|grep MonitorPlugin
  • Windows： wmic process where caption="MonitorPlugin.exe" get caption,commandline /value