Quick start
Where to start for an administrator who is responsible for operating a client cluster.
Who is this page for?
This is your first time working as a cluster administrator in the Shturval platform or you want to quickly find documentation for a specific operational task.
The documentation in this section describes working with the client cluster (workload cluster): cluster status, nodes, network, access, quotas, storage, backup, services and maintenance. Platform installation, infrastructure providers, and the management cluster are the tasks of the platform administrator; the general scheme is described on the Platform Architecture page.
How to work with the platform
Cluster administrators primarily work in the GUI. Navigation within a cluster — dashboard, namespaces, Import, and cluster sections — is described on the GUI overview page.
To work from the command line, use kubectl, kubeconfig, or the kubectl-shturval-plugin. Access options are on Access methods. For automation via HTTP API, see API overview and Swagger reference.
Typical cluster administrator tasks
Monitor the state of the cluster and respond to problems
- Start with the cluster dashboard: it shows deployment and update status, Kubernetes version, node groups, events, logs, certificates, metering, and service availability.
- To regularly check the status, use cluster status monitoring.
- When investigating incidents, look at logs in the platform, monitoring troubleshooting and logging troubleshooting.
Configure alerts and application monitoring
- For metrics, configure alert by metrics.
- For events in logs, configure log alerts.
- To ensure that alerts reach the team, configure alert recipients.
- For application metrics, use application monitoring.
Manage nodes and scaling
- For Control Plane and Worker groups, machine templates and autoscaling, use node groups.
- For single node operations: Cordon, Drain, removal, taints and diagnostics - see node page.
- For horizontal and vertical scaling of node groups, see node scaling.
- To configure the OS and runtime on nodes, use node configuration.
- To automatically recover unhealthy machines, configure MachineHealthCheck.
- To isolate workloads and control pod placement, use Taints, Tolerations and Affinity.
Operate Shturval v2 and hybrid clusters
- To add physical hosts and increase the group of nodes, use the Adding hosts to the cluster instruction.
- To reduce a group of Shturval v2 nodes, see Removing hosts from a cluster.
- To release a specific host, follow Release a specific host.
- For mixed clusters with provider nodes and Shturval v2, see Hybrid clusters.
Configure network and application access
- For incoming traffic and load balancing, configure Ingress controllers.
- For outgoing traffic, use Egress Gateway.
- To isolate pods and namespaces, configure network policies.
- For additional interfaces and Multus, use additional networks.
- In case of network incidents, use network troubleshooting.
Manage access and separation of responsibilities
- Assignments at the cluster and namespace level are made on the Cluster Access Control page.
- The general procedure for assigning rights to users and groups is in Assigning access rights.
- If you need to understand the composition of access sets and permissions, see Setting Permissions.
- For temporary access to pod diagnostics, use debug role.
Limit command and workload resources
- For CPU, memory and other resource limits in namespaces, configure Resource Quotas and LimitRanges.
- For CPU reservation at the cluster level, see CPU reservation in a cluster.
- To allocate separate node pools for namespaces or workload types, use Taints, Tolerations and Affinity.
Manage storage and protect data
- For storage classes, use StorageClasses.
- For persistent volumes, see PersistentVolumes.
- For clusters on OpenStack, VK Cloud, or Selectel providers — OpenStack CSI (Cinder).
- For volume claims and changes, use PersistentVolumeClaims.
- For volume snapshots, use snapshots.
- To choose between snapshots and backups, see Backups and Snapshots.
Set up backup and recovery
- First connect S3 storage for backups.
- For one-time and regular backups, use create backups.
- For disaster recovery or migration, see recovery from backup.
- In case of errors, use backup troubleshooting.
Install services and connect repositories
- To install Helm charts via the platform, API or
kubectl, use the instructions How to deploy a new service. - For third-party or internal Helm charts, connect your repository.
- For images from closed registries, connect private registry.
- To override Helm values, use CustomValues services.
- To work with platform CRDs, see Custom platform resources.
Upgrade and maintain the cluster
- To update Kubernetes use version transitions.
- Before updating, check nodes and machine templates.
- In case of update errors, use update troubleshooting.
- To maintain etcd, use defragmentation and ETCD backup.
- For TLS in the cluster, configure ClusterIssuer and Issuer, including Let’s Encrypt (ACME).
Where to start
- Open the cluster dashboard and check the status on the cluster status monitoring page.
- Set up basic alerts: by metrics, by logs and recipients.
- Check node groups, MachineHealthCheck rules and, if necessary, node configuration.
- Configure cluster access for users and teams.
- Connect S3 storage for backups and create backups using the Creating backups page.
- After the basic check, move on to specialized tasks: networks, quotas, storage or services.