Skip to content

ROADMAP.md

Latest commit

 

History

History
58 lines (53 loc) · 3.44 KB

ROADMAP.md

File metadata and controls

58 lines (53 loc) · 3.44 KB
Raw

Roadmap

GhostFleet 1.0.0 is feature-complete for its initial scope. This document lists what is implemented and what may come next. The how-and-why of the technical choices lives in ARCHITECTURE.md, TECH-STACK.md and DESIGN-DECISIONS.md.

Implemented

  • Core domain & API — SQLite-backed profiles (versioned), hypervisor connections (encrypted credentials), and deployments with effective-config snapshots and delete-keeps-history semantics; a REST API with an OpenAPI spec; optional password / API-key auth; and a Vue 3 web UI.
  • vSphere plugin — connection validation, placement enumeration (cluster / host / datastore / folder / resource pool), create/delete UEFI VMs, tag handling, and power operations. Deploy / scale-up / teardown are built on desired-vs-actual reconciliation and are retryable after partial failures. Includes an owner-aware cross-deployment conflict flow (adopt / clean / abort).
  • IPv6 PXE boot chain — dnsmasq (RA + stateful DHCPv6 + TFTP) + iPXE + an in-memory Alpine temp OS that hands off to the agent.
  • Calibrated data generation — fio-driven XFS fills with configurable size, compressibility and within-/cross-VM deduplicability, using deterministic hierarchical seeding, with live progress and throughput reporting.
  • Incremental runs & statistics — in-place, CBT-visible change and growth with no re-mkfs; run history, accurate write-phase throughput, and a stats dashboard with inline sparklines.
  • Restore validation (verify runs) — every fill/incremental stamps each disk with an identity marker and a per-file SHA-256 manifest; a verify run reads everything back and proves a backup restore round-tripped intact.
  • Discovery & adoption — restored ghost VMs (fresh MAC) PXE-boot into a report-only discovery mode, are inspected for GhostFleet identity, and can be adopted: rebind into their original deployment (restore-in-place) or into a new adopted deployment (restore-alongside) for verify/incrementals.
  • Schedules — timer-based run kickoff per deployment: recurring by interval or at a wall-clock time (with weekday filter and timezone), or one-shot (e.g. a teardown in 30 days). Persisted fire times survive controller restarts; missed fires are caught up once.
  • Operational hardening — idempotent deploys, run/job cancellation, static API keys, profile import/export, restart resilience, agent retry/resume, and fill boot resilience (wedged VMs are power-cycled; unbootable VMs fail alone).
  • Distribution — prebuilt container images on GHCR, built by a release workflow on every version tag, plus a pull-only compose stack; see INSTALL.md.

Ideas / not yet scheduled

  • Further hypervisor plugins (Proxmox VE, ESXi-native, Nutanix AHV, Hyper-V) — the plugin interface is platform-neutral by design.
  • Additional filesystems on generated disks (ext4, NTFS) and a raw-block mode (no filesystem) as a profile option.
  • Absolute-or-percent quantity inputs for change rate, growth rate and data amount (currently percentages).
  • Dynamic mid-run redistribution of the aggregate write-rate cap.
  • Hot-spot change distributions.
  • Export of run statistics (CSV / JSON) and a webhook on run completion.
  • Support for containers and physical machines as data-generation targets.
  • Packaging niceties: an OVA appliance, a cloud-init user-data example, and a multi-arch controller image.