Skip to main content

Container map metrics and status

Customizing the chart metrics

The property of the default chart provided by the WhaTap Kubernetes monitoring service's Container Map menu is the state of container or pod. The metric name is ConState (Container State). However, it is required to construct the map based on the criteria other than the default state. For example, it may be required to determine the states of each target based on the CPU usage (millicore), memory usage (%), network usage (byte), and such rather than the default state.

WhaTap provides the function to change the container's chart metrics. Changeable chart metrics values are the same ones that WhaTap Kubernetes collects.

Customizable chart metrics list

Note

The grouping settings (including the attribute fields and filtering options) on the screen are stored in the user browser. To initialize the stored settings, select Reset icon Reset.

Metrics threshold settings and state colors

The states of each target can be classified by color by setting the threshold values of the metrics given to the targets. Thresholds can be set when the units of metrics are for numbers (pct, bps, millicore).

The colors and meanings for states such as Normal, Warning, Critical, and such are as follows:

Target status color

StatusDescription
NormalDisplays the normal section (three divisions under Warning).
WarningIndicates it is within the Warning threshold (more than Warning to Critical or more).
CriticalIndicates it is within the Critical threshold (more than Critical).
DeadContainer exit state
WaitContainer startup state
PauseContainer's suspended state

Event-based container state

In Container Map, the container view provides a state value called EventConState. EventConState is the event-based container state, which represents the container state where the event is reflected. The event types reflected to EventConState are as follows:

EventLevel
Kubernetes EventWarning
WhaTap EventWarning, Critical
  • Kubernetes Event is the event data provided by Kubernetes. It is the same as the event data that can be found in the user's Kubernetes environment through the kubectl get events command.

  • WhaTap Event is set through the WhaTap event function.

EventConState is determined by the following methods:

StatusState valueCause
Critical50Critical event caused by the WhaTap microagent
Critical event caused by the WhaTap node agent
Toggle container state OOMKILLED
Warning30Warning event caused by Kubernetes
WhaTap microagent-caused Warning event
WhaTap node agent-caused Warning event
Container state RESTARTING transition
Dead23Transition of the container state to DEAD
Pause22Transition of the container state to PAUSE
Waiting21Transition of the container state to WAITING
Normal10Transition of the container state to Normal

Among the WhaTap events, there are events that have set Additional notifications when the event state is resolved.. Internally, two types of events are distinguished as follows:

  • Events that do not trace the state resolution: stateless events

  • Events that set additional notifications when the state is resolved: stateful events

The stateless and stateful events generated by the container are used to determine the EventConState state as follows:

  • stateless events: Reflects stateless events that occurred in the container within the last minute to the EventConState state.

  • stateful events: Reflects stateful events from occurrence to termination to the EventConState state.