Self-healing infrastructure platform
Operations
Basalt’s operational contract is built into the management plane: task tracking, metrics, alerts, maintenance drains, image-based OS updates, and first-boot setup all run through the same platform that reconciles infrastructure state.
Task & Job System
Every long-running operation is tracked as a task with progress, logs, and dependencies. Task
states include pending, executing, completed,
failed, timed_out, and cancelled.
The gateway sweeps timed-out tasks every 30 seconds while agents long-poll and stream progress back to the control plane. Operators can see what the platform is doing instead of reverse engineering state from host logs.
Built-in Metrics
Basalt records CPU, memory, disk I/O, and network I/O for every host and VM. Metrics are stored in partitioned timeseries tables with 30-day retention.
Day-one visibility does not require deploying an external monitoring stack before the first cluster can be operated. The built-in metrics system gives the gateway and operators the data needed for placement, troubleshooting, and capacity awareness.
Alerting
Alert rules support silence and resolve workflows inside the platform. The gateway sweeps unsilenced alerts every 60 seconds so actionable platform conditions remain visible.
Alerting is in-platform today. External delivery targets such as webhook, email, and Slack are on the roadmap rather than prerequisites for initial operations.
Maintenance Mode
Maintenance mode gracefully drains a host and suggests migration targets for workloads that need to move. The host status flag prevents new VM placement while maintenance is active.
The drain workflow is tied to VM lifecycle management in the compute control plane, so host service operations and workload placement share one source of truth.
OS Update Orchestration
Basalt uses bootc OCI image atomic swaps for host operating system updates. The agent
reports up_to_date, available, staged, and
error states along with version digests.
Image-based atomic upgrades make host updates observable and reversible in the same operational model as the rest of the platform.
Setup TUI
basalt-setup is the first-boot configurator. It runs as a Ratatui interactive TUI for
hands-on installation or as a headless JSON mode for automated environments.
The setup tool probes networking, writes configuration, installs CA trust, and validates gateway reachability. It is one binary with two modes rather than separate setup paths for manual and automated deployments.
Architecture Linkage
These operations are not bolted onto a separate stack. They are part of the same three-component Basalt architecture that drives manifest reconciliation and self-healing convergence.