When managing a cluster, you come across the terms Kubernetes (CSI, PV, PVC), networking (CNI, Cilium), monitoring and Shturval platforms. This page contains brief definitions. For platform, provider, and IAM installation terms, see Platform Administrator’s Glossary.
The cluster where the Shturval platform is deployed. ClusterManager, shturval-backend, shturval-frontend and other management components work here. The platform administrator works with this cluster.
Client Cluster
A cluster created through an application hosting platform. The cluster administrator manages the client cluster (nodes, services, backups, etc.).
Kubernetes standard for cluster lifecycle management. The platform uses for nodes: Machine, MachineDeployment, InfraMachineTemplate.
Machine
A CAPI resource that describes a cluster node. Created by CAPI controllers when adding nodes.
MachineDeployment
A CAPI resource that manages the Machine group. Analogous to Deployment for nodes: sets the desired number of nodes and a template.
InfraMachineTemplate
CAPI template for creating node infrastructure resources. Examples: VSphereMachineTemplate, OVirtMachineTemplate - depend on the provider.
Control Plane (CP)
Cluster management nodes. Launch the API server, etcd, scheduler, controller-manager.
Worker
Nodes for hosting applications. Kubelet, CNI, CSI work.
Infra Worker
A group of Worker nodes for system services (monitoring, logging, Ingress).
NCI (NodeConfigItem)
Platform object for applying configuration to nodes (sysctl, NTP, Container Runtime, packages, etc.). The shturval-node-config operator applies settings by selector. More details: Node Configuration (NCI).
MHC (MachineHealthCheck)
CAPI resource for automatic recovery of non-working nodes (replacement of Machine in case of failure).
Cluster Autoscaler
Component for automatic scaling of node groups based on load. See Node Scaling.
Cordon
Blocking the scheduling of new pods on a node without stopping those already running.
Drain
Safely evict pods from a node before servicing or deleting.
Node Conditions
Node states in Kubernetes (Ready, MemoryPressure, DiskPressure, etc.). See Conditions reference.
Hybrid Cluster
A cluster with nodes of different providers or infrastructure types. See Hybrid Clusters.
etcd
Kubernetes distributed key-value storage. Stores the state of the cluster.
Default limits and acceptable ranges for pods and PVCs in namespace (applies to new objects).
Taint
A label on a node that restricts scheduling of pods without an appropriate Toleration.
Toleration
The permission in the pod specification is to be scheduled on nodes with the specified Taint.
Affinity
Rules for node preference (nodeAffinity) or placement relative to other pods (podAffinity / podAntiAffinity). More details: Taints, Tolerations and Affinity.
HPA (Horizontal Pod Autoscaler)
Automatically changes the number of Deployment or StatefulSet replicas based on CPU, Memory, or custom metrics. See HPA on deployment.
VPA (Vertical Pod Autoscaler)
Automatically selects container CPU/Memory requests and limits from historical usage. In the platform, VPA is configured for Deployment and StatefulSet. See VPA on deployment.
Metrics system (analogous to Prometheus). vminsert, VMSingle, VMAlertManager - components.
VictoriaLogs
Log storage system in the platform. An instance for each cluster is deployed in the management cluster - VLSingle; collection - shturval-log-collector.
VLogs
Brief designation of the VictoriaLogs log subsystem in the GUI and documentation.
Grafana
Metric visualization tool. In the cluster GUI, access is through the Statistics button on the dashboard.