v1.2.5

[1.2.5] - 2026-10-03

Fixed

• Online snapshots failed at the filesystem freeze, and the agent claimed it had frozen. guestkitd runs as zyvor-agent with an empty capability set, but the QGA handlers guest-fsfreeze-freeze/-thaw ran fsfreeze themselves, so every freeze failed with Operation not permitted and KubeVirt failed the snapshot of any VM with PVC-backed disks (measured on Ubuntu 24.04 under KubeVirt 1.9: a freshly installed agent cannot freeze; with CAP_SYS_ADMIN it can). The unit file already says privileged operations run in guestkitd-exec; freeze and thaw now do: a new fsfreeze action in the helper (freeze/thaw only, always /) and executor_ipc::fsfreeze() for the agent, falling back to an in-process call only when no helper is running (agent run as root). Also, snapshot_hooks::freeze_filesystems/thaw_filesystems treated a successfully spawned fsfreeze as success (status().is_ok()), so snapshot.prepare reported fs_frozen: true while nothing was frozen; they now use the same checked path. Covered by a test that starts the real guestkitd-exec with a fake fsfreeze. Requires the helper service (zyvor-guest-agent-exec.service, enabled by the packages) to be running.
• Linux agent never connected to the hypervisor channel on stock distros. The virtio port (/dev/virtio-ports/org.qemu.guest_agent.0) is root:root 0600 but guestkit-agent.service runs as zyvor-agent, so the agent started, logged failed to open virtio channel, and KubeVirt reported no guest agent. The DEB, RPM and tarball now ship 60-zyvor-guest-agent.rules (group zyvor-agent, mode 0660 for org.qemu.guest_agent.0 and com.zyvor.guestkit.0) and reload/trigger udev on install. Verified on Ubuntu 24.04 under KubeVirt 1.9: with the rule the agent connects (agentConnected), without it it does not.
• Concurrent NBD allocate+connect — NbdDevice::connect holds a cross-process flock (/run/lock/guestkit-nbd.lock, fallback /tmp) across free-device selection and qemu-nbd -c, and retries the next free index on failure. Fixes races where two mounts claimed the same /dev/nbdN (fluxvm#104: /dev/nbd0p1 already mounted, wrong guest-agent token). new() no longer claims a device index before connect.

Added

• Offline inject on run_migrate_repair — optional inject_json appends hostname, network files, users, services, first-boot scripts, cloud-init user-data, Active Directory rejoin, Windows KMS reactivation, and RDP enable to the repair plan. Empty or "null" is a no-op. guestkit migrate-repair has no inject flag; h2kvm passes the JSON from Python.
• Live guest fix helpers — live_fix_commands() returns shell lines (regenerate initramfs, update GRUB, optionally remove open-vm-tools). run_live_plan(commands, dry_run=False) runs those lines on the machine where Python is executing. It does not SSH.
• Docs: README is now a landing page — the detailed sections moved into docs/ (quick-start, who-does-what, capabilities, oss-vs-enterprise, platform-layout, repository-layout, documentation-map, ...); every earlier README anchor still resolves.

v1.2.4

[1.2.4] - 2026-09-18

Changed

• Docs — install instructions use the GitHub Release and pip install zyvor-guestkit. GHCR examples pin v1.2.4. Serial console without grubby is documented. crates.io guestkit is still 0.3.2.

v1.2.3

[1.2.3] - 2026-09-18

Fixed

• Serial console without grubby — Photon and other guests have no grubby. Kernel console args are written into the bootloader the guest already uses (grub2, BLS, syslinux, extlinux, zipl) instead of failing first-boot with exit 127.
• XFS remount — pick a valid remount option set after norecovery inspect mounts.
• Guest root resolution — path resolve uses the real guest /, not sda1/sda2.
• Windows agent inject — mount the Windows root read-write for offline guestkitd inject.

Changed

• PyPI — Python distribution is zyvor-guestkit, owned by pypi.org/user/zyvor. hypersdk-guestkit is retired. Import name stays guestkit. Wheel artifact: zyvor_guestkit-*.whl.
• GHCR registry — container images publish/pull from ghcr.io/zyvorai/{zyvor-ui,zyvor-api,guestkit-worker} (was ghcr.io/hypersdk).
• Docs — customer wording is now users; CE vs Enterprise licensing is spelled out; docs site is Docusaurus.

Removed

• Zyvor company terms and the distribution acceptance gate.

v1.2.2

[1.2.2] - 2026-09-06

Added

• Web Image Vault TUI parity — OSS deploy/ui renders inspect inventory tabs, Assurance (doctor / migration plan / passport / repair preview + gated apply), Profiles/Issues, and read-only Files browse (guestkit.explore). Docs: using-the-dashboard.md.
• POST /api/v1/vms/:id/profile and POST /api/v1/vms/:id/explore — enqueue existing/new worker ops; repair-plan accepts dry_run query (default true).
• OpenAPI documents profile, explore, and repair dry_run.

Changed

• Inspect worker summary — larger package/service samples (200/100) for UI lists.

v1.0.1

[1.0.1] - 2026-08-15

Fixed

• guestkit-worker's Docker release build failed for every release past
  the 0.x line — crates/guestkit-worker/Cargo.toml pinned its own
  guestkit path dependency to version = "0.3.3"; the worker's
  Dockerfile never copies a Cargo.lock in, so each build freshly
  resolves dependencies and cargo enforces the version requirement even
  against a path dependency — 1.0.0 didn't satisfy ^0.3.3 and the
  Publish GHCR Release Images job in v1.0.0's release run failed.
  Dropped the version pin (path-only, matching zyvor-api and
  zyvor-guest-agent's existing pattern for the same in-workspace
  dependency).

v1.0.0

[1.0.0] - 2026-08-15

Added

• GitHub Action for the Passport CI gate (action.yml) — reusable
  composite action wrapping doctor → migrate-plan → passport emit → passport verify as a single CI step; installs a checksum-verified
  release binary, no build step. Dogfooded against a real disk image by
  .github/workflows/passport-gate-demo.yml on every change. See
  docs/devops/01-passport-ci-gate.md.
• Native OpenAI tool-calling for the AI copilot (src/ai/rig_tools.rs)
  — rig-core AgentBuilder/multi_turn with real JSON-schema tool
  definitions, replacing regex/JSON-scraped completion text for OpenAI.
  xAI/Anthropic/Ollama still use the original text-instructed loop.
• Cross-run AI memory (src/ai/memory.rs) — a repeated doctor --ai/
  migrate-plan --ai run against the same VM folds a summary of prior
  findings into the query, capped at the last 20 runs.
  GUESTKIT_AI_MEMORY_DIR/GUESTKIT_AI_MEMORY=0 to relocate/disable.
• MCP server for the AI copilot (src/ai/mcp.rs, --features mcp) —
  guestkit mcp-serve <disk> [--target <target>] exposes the same 6
  read-only evidence tools over stdio to Claude Desktop / other MCP
  hosts, independent of guestkit's own agent loop.
• guestkit fleet wave-plan (src/fleet/wave.rs) — orders a fleet's
  disk images into dependency-aware migration waves (DB-role priority +
  NFS storage-dependency edges, Kahn's-algorithm topological sort,
  cycles reported rather than dropped or arbitrarily ordered).
• guestkit fleet watch (src/fleet/baseline.rs) — scheduled drift
  monitoring: diffs each VM's current evidence against a stored golden
  baseline (first run establishes it), --fail-on-drift for pipeline
  gating. Includes a Kubernetes CronJob template
  (deploy/helm/zyvor/templates/fleet-drift-watch-cronjob.yaml) as the
  reference scheduled-invocation path.
• Helm chart CI — ci.yml's new helm-chart job runs helm lint and
  helm template against deploy/helm/zyvor, which previously had zero
  CI coverage, across default values, every optional PVC-backed feature
  enabled at once, hostPath-backed persistence, and all three real
  deployment overlays (values-ci.yaml, values-k3s.yaml,
  values-prod.yaml).
• Manual-dispatch workflow for NBD-dependent tests
  (self-hosted-nbd-tests.yml) — runs the full test suite with none of
  ci.yml's NBD skips, for self-hosted runners with working loop/NBD
  support. workflow_dispatch only, deliberately never wired to
  pull_request/push.

Fixed

• k3s E2E's zyvor-api pod crash-looped on every run — the Helm chart's
  default zyvorApi.agentMtls.enabled: true requires AGENT_BOOTSTRAP_TOKEN
  (zyvor-api refuses to start otherwise: "AGENT_MTLS_BIND_ADDR is set but
  AGENT_BOOTSTRAP_TOKEN is unset"), but the Deployment template only ever
  wired that env var — and the Secret holding it — inside the
  zyvorApi.auth.enabled block. values-ci.yaml (mTLS on, full auth off)
  hit exactly that gap; values-prod.yaml masked it by having both auth and
  a token on together, and values-k3s.yaml worked around it by disabling
  mTLS outright (its own comment already described the bug). Decoupled the
  zyvor-api-auth Secret and AGENT_BOOTSTRAP_TOKEN env var from
  auth.enabled — gated only on the token itself being set, matching what
  zyvor-api's own config validation actually requires — and set a
  CI-only placeholder token in values-ci.yaml so the E2E job now exercises
  the mTLS path instead of crash-looping. ci.yml's Helm Chart job now also
  helm templates all three real overlays (values-ci.yaml,
  values-k3s.yaml, values-prod.yaml) so a rendering break here is caught
  without needing a live k3s cluster.
• k3s E2E multi-round debugging — the job was failing on every run;
  fixing it required peeling through several layers, each masking the
  next:
  • poll_job (deploy/scripts/e2e-smoke.sh) only recognized
    "completed" as terminal, so a "failed" job status looked
    identical to "still pending" for the full 5-minute poll budget, and
    the worker's own error message (live_status.error) was never
    printed. Now treats failed/cancelled/timeout as terminal and
    prints the error immediately.
  • install-k3s-ubuntu.sh never got the loop/NBD device setup
    ci.yml needed earlier this session (guestkit-worker's pod
    bind-mounts the host's /dev, so the same root:disk-0660-node /
    EACCES-looks-like-timeout issue applies here too). Added it —
    confirmed live afterward: inspect/doctor both complete in one
    poll with a real bootability score. This specific k3s stack path
    does not hit the deeper NBD-attach limitation ci.yml's plain
    cargo test job still has to skip around.
  • curl -sf swallows the response body on any non-2xx status, so the
    next failure (provision) looked like an empty response
    (Expecting value: line 1 column 1) instead of a real API error.
    Added a curl_or_die helper (splits HTTP status from body via
    curl's -w, prints both on failure).
  • That revealed a real HTTP 500: "No operating system found in disk image" from provision (POST /vms/{id}/provision, which mounts
    the disk synchronously in zyvor-api's own process), on the same
    image doctor had just inspected successfully. Suspected (and
    partially fixed) an unawaited async migration-plan job racing
    provision's own mount via NbdDevice::find_available_device
    (src/disk/nbd.rs) — that function does check device availability
    and connect as two separate, unlocked steps with no cross-process
    coordination, a real bug now flagged with a code comment — but
    serializing migration-plan before provision did not fix it,
    disproving the race as this failure's cause.
  • provision_vm's .map_err(|e| ApiError::internal(e.to_string()))
    (crates/zyvor-api/src/routes/vms.rs) only shows anyhow's outermost
    .context() layer via plain Display. Changed to format!("{e:#}")
    (alternate Display, full chain) for provision_vm's three
    guestkit/export::kubevirt-derived map_err calls — correct and
    worth keeping, but the next run's error was still byte-identical to
    before, because there was no chain to reveal: mount_all_ro
    (src/cli/commands/mod.rs) returns Option<String>, not
    Result<String>, and .context("No operating system found in disk image") on a None (anyhow's Context impl for Option) produces
    an error with no wrapped source at all — the context message
    genuinely is the entire error. Left the other ~90
    .map_err(|e| ApiError::internal(e.to_string())) sites in the crate
    alone; most wrap simple error types (serde_json, std::io) where
    .to_string() isn't lossy.
  • The real swallowed information was one layer further down:
    mount_all_ro calls g.inspect_os().unwrap_or_default() —
    inspect_os() failing for any reason (guestfs launch issue, mount
    error, permission problem) collapses to the identical empty-roots
    None as "genuinely no OS found," discarding whatever inspect_os's
    real error was before it could reach any context message. Changed to
    log the real error (log::warn!) before discarding it. Didn't widen
    mount_all_ro's Option<String> return type to Result — it's used
    across 9 files where callers only ever branch on Some/None, and
    that ripple is out of scope for this investigation. Widened the
    default EnvFilter (was "nothing enabled" without RUST_LOG set,
    now falls back to warn globally) so mount_all_ro's new
    log::warn! — and any other log::*! from guestkit's dependency
    graph — actually reaches the pod's logs. First attempt at this also
    added an explicit tracing_log::LogTracer::init() call, reasoning
    that zyvor-api only sets up a tracing subscriber and guestkit
    logs through the plain log facade — wrong, and a real
    regression: tracing-subscriber's "tracing-log" feature (on by
    default) already bridges log into the subscriber as part of
    .init(), so the explicit call double-registered the global log
    logger and panicked at startup with SetLoggerError, crash-looping
    zyvor-api again. Confirmed via kubectl logs from the next E2E
    run (once the "dump pod logs on failure" step below existed to
    capture it) and reproduced in an isolated 10-line binary before
    re-pushing — removed the explicit LogTracer::init() call and the
    now-unneeded direct tracing-log dependency; the isolated repro
    confirmed log::warn! still reaches the subscriber correctly
    without it.
  • Also: nothing in k3s-e2e.yml ever captured pod logs on failure —
    every fix in this list up to this point was diagnosed purely from
    HTTP response bodies, each requiring a full ~20-40min re-run just to
    test. Added a failure-only step dumping kubectl logs (all
    containers, prefixed by pod) for every deployed component, plus
    get pods -o wide and describe pods.
  • That finally showed it: no panic, no error, no mount_all_ro
    warning — inspect_os() genuinely returned an empty root list.
    validate_root_partition/validate_initrd_boot_partition
    (src/guestfs/inspect.rs) treat mount/extraction failures as
    "not a valid root" by design, not as errors — a real mount
    failure and "genuinely no OS" are indistinguishable at that layer
    on purpose (LVM volumes on read-only NBD devices can legitimately
    fail to mount for benign reasons). validate_initrd_boot_partition
    is the cirros-cloud-image path — root filesystem lives inside the
    initrd, not on a directly-mountable partition — and shells out to
    zcat <initrd> | cpio -t to look inside it.
    crates/zyvor-api/Dockerfile never installed cpio (or gzip),
    unlike crates/guestkit-worker/Dockerfile's otherwise-ident...

v0.3.20

[0.3.20] - 2026-08-07

Added

• DevOps runbooks — docs/devops/ Passport CI gate, offline repair worker,
  air-gap packages/VirtIO, fleet analyze, cutover weekend, failure triage,
  cloud disk sources (S3/GCS/Azure), forensic IR, SBOM/inventory CI.
• GitHub Wiki — operator cheat sheets (Passport, day-0, packages, env,
  TUI, KubeVirt/GCF/agent) linked from README / docs INDEX.
• GUESTKIT_PACKAGE_MIRROR — HTTP fallback via curl/wget when host
  dnf/apt-get is missing or fails (comma-separated bases; optional
  {name}/{ext} templates). Helps macOS hosts stage PackageInstall.
• Domain-leave first-boot RunOnce — windows-domain-leave stages
  GuestKitDomainLeave RunOnce (Add-Computer -WorkGroupName) in addition
  to Tcpip/Winlogon markers (DC computer-account delete still needs live AD).
• Worker performance + migration profiles — guestkit.profile jobs run
  the same CLI InspectionProfile implementations as guestkit profile.
• Offline ServiceOperation / CommandExec staging — enable/disable via
  systemd wants Symlink/FileDelete; start/restart and other commands stage
  guestkit-firstboot-live.service when chroot cannot run them.
• UEFI-aware fix-grub --force — detects ESP under the guest root and
  runs grub-install --target=x86_64-efi|arm64-efi --efi-directory=… --no-nvram --removable (BIOS path unchanged).
• Windows AES/RC4 SAM NT-hash write — rescue -o reset-password --password
  reconstructs the SYSKEY bootkey from SYSTEM LSA class names, derives the
  hashed bootkey from SAM F, and writes an AES-128-CBC (or legacy RC4)
  encrypted NT hash into the user V blob. Falls back to SAM blank + RunOnce
  net user if SYSTEM/bootkey/crypto fails.
• PackageInstall host fetch — with GUESTKIT_PACKAGE_FETCH=1, offline
  plan apply downloads missing .rpm/.deb on the host (dnf download /
  yumdownloader / apt-get download) into GUESTKIT_PACKAGE_CACHE or
  ~/.cache/guestkit/packages, then stages the first-boot oneshot as before.
• Offline GRUB repair (fix-grub) — rescue -o fix-grub bind-mounts
  proc/sys/dev and runs chroot grub2-mkconfig / grub-mkconfig /
  update-grub; --force also attempts grub-install onto the NBD device
  (BIOS) or EFI removable path when an ESP is present;
  if chroot mkconfig fails, stages guestkit-firstboot-grub.service.
  --export-plan writes the first-boot FileWrite/Symlink ops. check-grub
  remains diagnose-only.
• System Reserved / ESP detection — offline Windows evidence probes
  non-OS NTFS/FAT volumes for bootmgr + BCD (legacy System Reserved) or
  EFI Microsoft Boot (ESP). Surfaces on windows.system_reserved, promotes
  bcd_store_found / bootmgr_found, fixes esp_present (no longer aliased
  to bootmgr). Boot check BOOT-014, migration MIG-W-011, Passport
  flags system_reserved_layout + bcd_store_found.
• Windows driver/hotfix migration diagnostics — offline HotFix registry +
  $NtUninstall* / $hf_mig$ / CBS.log tail; VirtIO .sys presence on
  WindowsDriverEntry.sys_present; BCD UTF-16 probe for testsigning /
  nointegritychecks. Migration MIG-W-012 (hotfixes/servicing),
  MIG-W-013 (VirtIO files); Passport hotfix_count /
  hf_mig_present / driver_signature_enforcement. Hive paths resolve via
  guestfs mount root.
• Offline activation / ghost-NIC depth — SOFTWARE ProductId/EditionID/
  DigitalProductId + oeminfo.ini → windows.activation (OEM/Retail/Volume);
  SYSTEM Enum\PCI remnant/problem NICs → ghost_nics; Tcpip static
  interfaces → static_nic_configs. Enriches MIG-W-006/007/008; Passport
  activation_channel, ghost_nic_count, static_nic_count.
• Offline BitLocker / VSS enrichment — BitLockerStatus\BootStatus (On →
  hard block), FVE/$BitLocker/fvevol artifacts (offline_uncertain warning),
  VSS+swprv services + System Volume Information inference. Fills
  windows.bitlocker / windows.vss for MIG-W-005/009; Passport
  bitlocker_uncertain.
• Day-0 plan/rescue depth — windows-dhcp / windows-dns / linux-hostname
  profiles; rescue enable-rdp / enable-winrm / set-timezone; Windows
  set-hostname applies registry day-0 plan (was Linux /etc/hostname).
• Cutover Passport signed-enterprise workflows — passport keygen (Ed25519
  seed + pubkey); emit --issuer / --expires-hours; verify --trust-keys
  allowlist + --max-age-hours freshness gate (signing/verify need agent).
• Production Helm — values-prod.yaml: PVC-backed Postgres/Redis/MinIO
  (eval still emptyDir); Ingress TLS + cert-manager annotations; pinned
  GHCR v0.3.19 images; nightly image-vault backup CronJob + backup PVC.
• Guest Control Fabric poll telemetry — airgap reconciler records per-method
  latency + transport attempts; Redis fleet rollup; GET .../guest/poll-telemetry
  (VM + fleet); guest/status exposes lastPoll / telemetryMode.
• Fleet analyze performance — parallel --jobs / GUESTKIT_FLEET_JOBS
  (default min(4, CPUs)); evidence-cache hit skips remount.
• Cloud disk source depth — persistent ~/.cache/guestkit/cloud pulls;
  S3 GUESTKIT_S3_ENDPOINT/AWS_ENDPOINT_URL; azure:// URIs; GCS
  gcloud storage fallback; CI recipe scripts/ci-cloud-disk-sources.sh.
• Offline heuristic remediations + linux-grub — systemctl enable/disable
  → Symlink/FileDelete; fail2ban/auditd/chrony/apparmor/sshd enable offline;
  ufw default deny FileEdit; day-0 linux-grub (--grub-timeout /
  --grub-cmdline) for /etc/default/grub.
• Offline PackageInstall staging — when GUESTKIT_PACKAGE_CACHE (or
  host_cache) holds matching .rpm/.deb, offline plan apply stages
  packages + a first-boot systemd oneshot instead of skipping; optional
  GUESTKIT_PACKAGE_FETCH=1 downloads missing packages on the host first;
  live install unchanged.
• Windows offline password set — rescue -o reset-password --password
  prefers AES/RC4 SAM NT-hash write via SYSKEY; falls back to SAM blank +
  HKLM RunOnce net user for first boot; omit --password to blank only.

Changed

• Docs — Roadmap parked list cleared; CLI / quick-reference / feature guide /
  fix-plans updated for AES SAM passwords, fix-grub, and
  GUESTKIT_PACKAGE_FETCH offline staging.

v0.3.19

[0.3.19] - 2026-08-06

Added

• Cutover Passport — guestkit passport emit|verify: versioned CI-gateable
  assurance artifact (evidence digest, boot/migration scores, FixPlan digest,
  Windows BitLocker hard-block + windows_offline_ready, optional live
  attestation via agent-proxy, optional Ed25519 sign). Suite handoff points to
  HyperSDK (export) + hyper2kvm (convert/deploy). Web: POST /vms/:id/passport
  • dock download. Worker op guestkit.passport.
• plan generate -p windows-domain-leave — offline domain→workgroup
  markers (--workgroup, default WORKGROUP).
• plan generate -p windows-timezone — offline TimeZoneKeyName
  (--timezone).
• plan generate -p windows-static-ip — offline static IPv4 on a known
  interface GUID (--interface-guid --ip --mask [--gateway] [--dns]).

v0.3.18

[0.3.18] - 2026-08-06

Added

• plan generate --profile windows-hostname — offline ComputerName + Tcpip
  Hostname / NV Hostname (--hostname required). Apply with --skip-backup.
• plan generate --profile windows-winrm — WinRM Automatic +
  WINRM-HTTP-In-TCP firewall rule. Apply with --skip-backup.
• Symlink / FileDelete plan ops — offline guestfs ln_sf / rm
  (used by hardened linux-ssh).
• plan generate -p linux-ssh --user + --key / --key-file — inject
  authorized_keys into the enable plan.
• Windows rescue -o reset-password — offline SAM blank (chntpw-style)
  via registry-write / libhivex; clears password so interactive logon works.
• rescue --export-plan PLAN.yaml — emit a reviewable FixPlan for
  enable-ssh / inject-ssh-key / set-hostname / reset-password / fix-fstab.
• Offline DriverInject — apply uses host_dir / GUESTKIT_VIRTIO_WIN
  • inject_windows_driver_dir when built with registry-write,agent.
• migrate-repair --virtio-win DIR — wires VirtIO host tree into
  migration repair DriverInject (same as $GUESTKIT_VIRTIO_WIN).
• Heuristic offline remediations — firewalld enable → Symlink, ufw →
  conf edit, more sshd FileEdits; preview tags live-only ops as offline-skip.

Fixed

• linux-ssh plan fidelity — wants enable via Symlink (not CommandExec ln), removes /etc/ssh/sshd_not_to_be_run, matches rescue enable path.
• from_security_profile naming — plan profile/tags follow the inspect
  profile name (not hardcoded "security").
• rescue check-grub — diagnose-only rename (fix-grub kept as alias).

v0.3.17

[0.3.17] - 2026-08-06

Added

• plan generate --profile linux-ssh — offline Linux SSH enablement
  (systemd ssh/sshd wants symlink + /etc/ssh/sshd_config.d/99-guestkit.conf
  with PubkeyAuthentication yes). Apply with --skip-backup.
• plan FileWrite operation — create/overwrite a guest file offline.
• rescue inject-ssh-key — --user + --key / --key-file appends to
  authorized_keys.
• rescue set-hostname — --hostname writes /etc/hostname and patches
  /etc/hosts.

Fixed

• rescue enable-ssh — actually creates the systemd wants symlink and
  writes an sshd drop-in (previously only printed a manual systemctl note);
  write drop-in before unit enable; prefer real wants dirs / relative
  symlinks / ln -sfn when guestfs ln_sf rejects unit paths.
• Windows guest-fsfreeze-freeze/thaw — route to VSS marker shadows
  instead of the Linux fsfreeze binary so KubeVirt quiesced snapshots work
  on Windows guests.