Skip to content

Releases

pvewarden 0.1.7

Choose a tag to compare

@github-actions github-actions released this 04 Nov 15:15
  • pvewarden tells you about new versions. The server asks GitHub's release list once a day and
    check_updates reports a newer pvewarden to the agent, including whether a release since yours
    fixes a security problem, with the upgrade command. doctor checks it too. It only tells: the
    upgrade stays pvewarden-setup upgrade on the host. The request sends nothing about the
    installation; turn it off with check_new_version = false.
  • check_updates now returns {guests: [...], pvewarden: {...}} instead of a bare list of guests.

pvewarden 0.1.6

Choose a tag to compare

@github-actions github-actions released this 02 Nov 20:51

From a code review; no new features.

  • Upgrading needs 0.1.3 or newer. The code that migrated installs from 0.1.0 to 0.1.2 is gone
    (shared guest SSH key, unrecorded ownership of the pool, role and user, the old enroll.conf, the
    pip-installed server, guest clients without services). To upgrade an older install: uninstall
    and install again.
  • Fix: the install refused an agent name with spaces or dots only after the container existed (the
    config could not hold it). It is now checked when it is asked, like pvewarden-setup token does.
  • Fix: the install left umask 077 in effect after writing the server config, so the host's files
    written after it (enroll.conf, /opt/pvewarden) were readable by root only.
  • pvewarden-setup renew-cert turns HTTPS on with the new pvewarden config enable-tls in the
    container, which validates the config before writing it, instead of sed. The certificate is now
    in place before the config names it.
  • The upgrade's question about enrolling the guests again says what it does today: it updates the
    guest client (each guest gets a fresh key; permissions and automatic updates stay).
  • Docs are tool-neutral: docs/wireguard.md became docs/vpn.md, a guide for any VPN with WireGuard
    as one example; docs/clients.md describes proxies that add credentials in general. The service
    no longer orders itself after wg-quick@wg0: it retries every 10 s until its address exists,
    whatever the VPN.
  • Smaller code: shared helpers in pvewarden-setup (installing the server code, restarting the
    service, the MCP address, the install check) and in the server (snapshot confirmations, the
    update check, HTTP responses).

pvewarden 0.1.5

Choose a tag to compare

@github-actions github-actions released this 02 Nov 12:15

Clearer install, from the first production install:

  • The root password of the server container is described as what it is: console and pct enter
    only. Debian refuses root passwords over SSH; the wizard now says so and points to SSH keys.
  • Reaching the server through a VPN: the wizard explains the two addresses (LAN for the guests, VPN
    for the agents). A VPN address that is not up yet at the end of the install is the next step, not
    a failure: the summary says what to do, and the server starts by itself within 10 s once the
    address exists. New guide: docs/wireguard.md (also in Czech).
  • pvewarden-setup renew-cert switches an install made with plain HTTP to HTTPS on the same address
    (certificate signed by the Proxmox CA; tokens stay). With a VPN address, the wizard now defaults
    to HTTPS too: some MCP clients (NanoClaw) refuse a remote server over plain HTTP.
  • Enrolling: VMs whose QEMU guest agent does not answer are no longer offered (and skipped with
    --yes); the wizard lists them with the steps to add the agent, including the power cycle.
  • New pvewarden-setup doctor [-v]: the server's self-check from the host, without runuser and
    full paths. "run as root" now says sudo -i.
  • Automatic updates: the wizard says what on and off mean; doctor lists the guests where they are
    on as one [INFO] line with how to turn them off, instead of a note on every guest.
  • Docs: checking the downloaded script is optional (pvewarden-setup checks what it downloads);
    the upgrade steps no longer require it.

pvewarden 0.1.4

Choose a tag to compare

@github-actions github-actions released this 02 Nov 08:31
  • Fix: pvewarden-setup upgrade installed the version recorded in /etc/pvewarden/setup.conf
    (the one installed last) instead of its own. 0.1.3 therefore switched to the newest setup, which
    installed the old version again; the downgrade blamed on an old script in 0.1.3 had this cause.
  • Upgrading: from 0.1.3, pvewarden-setup upgrade switches to 0.1.4's setup by itself. From
    0.1.2 or older, download the current pvewarden-setup first (docs/install.md).

pvewarden 0.1.3

Choose a tag to compare

@github-actions github-actions released this 01 Nov 23:28
  • pvewarden-setup upgrade first switches to the newest release's pvewarden-setup (checksum
    checked) and lets it do the upgrade. Before, the installed script installed its own version, so
    an old script could downgrade the server: 0.1.0's setup put 0.1.0 back over 0.1.2.
  • The upgrade refuses to install an older version than the server runs, before changing anything
    (--allow-downgrade to do it anyway).
  • Upgrading from 0.1.2 or older: download the current pvewarden-setup first (docs/install.md).

pvewarden 0.1.2

Choose a tag to compare

@github-actions github-actions released this 01 Nov 23:07

Security

  • Every guest has its own SSH key, made on the server when the guest is enrolled. Before, guests
    shared one key and the server connected to whatever address a guest reported, so a compromised
    guest with the update tag could report another guest's address and have updates run there
    without a snapshot and without that guest's tags. Now the server picks the key by VMID and the
    other guest refuses it. Guest host keys are remembered per VMID too.
  • Enrolling switches a guest to its new key in checked steps and never leaves it half switched: if a
    step fails, the server refuses the guest until it is enrolled again. Re-enrolling rotates the key.
  • Upgrading: pvewarden-setup upgrade enrolls the guests again and then removes the old shared
    key. Guests that are not enrolled again are refused ("has no pvewarden key of its own yet").
  • A shared [ssh] key is still possible, but only for guests whose address is fixed in the config.
  • New pvewarden key new|verify|activate|remove|list for manual installs.

One change per guest at a time

  • Every change to a guest (snapshot, update, waiting for an update, reboot, rollback, snapshot
    deletion) holds a lock for that guest; a second one is refused at once. Before, two updates of one
    guest called at the same time took two snapshots, and the second removed the first one's record of
    the running update, which unblocked reboot and rollback while dpkg ran.
  • A running update is recorded before anything changes, with a random run id, and outlives a server
    restart. Only its own run removes it. The guest names each run, and update_result lifts the
    block only when the result is provably from that run, or when provably nothing ran
    (never_started). When pvewarden cannot tell (e.g. the update process vanished during dpkg), the
    guest stays blocked and a human clears it after checking: pvewarden pending list|clear <vmid>.
  • A lost SSH connection during an update no longer counts as a failed update: the guest stays
    blocked and update_result picks it up.
  • Writes of that record are fsynced together with its directory (and the directory's creation), so
    on file systems that honour fsync it survives a power loss.
  • Enrolling a guest takes its lock in each step, and a new key is refused while an update's result
    is unknown.

Audit log

  • A call that is cancelled (client gone, server stopping) still gets its tool_result record
    (cancelled).
  • Every tool call gets an id and two records: tool_call (agent, IP, arguments) before it runs and
    tool_result (ok, error with the reason, or asked_user, and the duration) after. Records
    written in between carry the same id. Before, errors and refusals had no outcome record.
  • A tool that changes something does not run if its tool_call record cannot be written.
  • Refused requests are summarised per address and minute (http_denied_summary with a count)
    instead of one line each, so a flood of them cannot fill the disk.
  • The HTTP layer no longer reads every request body into memory for the audit log.

Limits

  • Requests larger than 1 MiB are refused (413), also when they are streamed.
  • More than 16 MiB of output from one guest command is refused; the run log stays as small. Output
    is read in blocks, so one endless line from a guest cannot break reading.
  • The server rotates its audit log weekly (a year kept, logrotate) and removes run logs older than
    180 days (systemd-tmpfiles).

Supply chain

  • requirements.lock pins every dependency with SHA-256 hashes. The server installs them with
    --require-hashes --only-binary=:all:: no unpinned build tools, no pip upgrade from the network.
    pvewarden itself is copied from the checked source instead of being built.
  • Release files get a signed build provenance attestation
    (gh attestation verify <file> -R matousfiser/pvewarden).

Rollback

  • After every rollback pvewarden waits until the guest answers again and reports its state, also
    when the guest kept running (before, only when it had to start it).
  • The security model documents what a snapshot is (disk only, crash-consistent; file system
    consistent with the guest agent's freeze) and what that means for databases.
  • Open, accepted as lower priority for now: snapshots are recognised as pvewarden's by their name
    prefix alone (rollback and deletion still need the user's confirmation). Planned for 0.1.3.

Installer

  • A pool, role or user with pvewarden's names that the installer did not create is no longer taken
    over silently: the install stops before any change, or takes them over only when you say so
    (dialog, or PVEWARDEN_ADOPT_EXISTING=yes). Before, it rewrote an existing Warden role's
    privileges and uninstall deleted such objects. uninstall now deletes only what the installer
    created or took over (/etc/pvewarden/owned.conf); upgrades record the objects of older installs.
  • Upgrading an install from before 0.1.2 no longer records the pool, role and user as created by
    pvewarden: older installers may have taken over existing ones. The upgrade asks (default no;
    PVEWARDEN_ADOPT_EXISTING=yes with --yes). uninstall always removes pvewarden's API token,
    also from a user it leaves in place, and prints how to remove what it left.
  • When a step fails, pvewarden-setup says so instead of stopping without a word.
  • When a guest refuses the server's new key because the server's address changed (DHCP),
    pvewarden-enroll says so; docs/troubleshooting.md has the steps to move to a new address.

pvewarden 0.1.1

Choose a tag to compare

@github-actions github-actions released this 31 Oct 23:33
  • pvewarden-setup: the install summary, connect and token output are laid out in clear
    sections that fit 80 columns. Long commands start at the left edge, so they wrap cleanly and copy
    well. Colors appear on a terminal, never with NO_COLOR or when the output is redirected.
  • The connection steps and the token output describe any MCP client first; Claude Code is one
    example.

pvewarden 0.1.0

Choose a tag to compare

@github-actions github-actions released this 31 Oct 22:58

First public release.

Server and tools

  • MCP server over streamable HTTP with one bearer token per agent, or over stdio.
  • Tools:
    • reading: list_guests, check_updates, check_services, list_snapshots,
    • updating: snapshot_guest, apply_updates, update_result, reboot_guest,
    • snapshots: rollback_snapshot, delete_snapshot, delete_old_snapshots.
  • Policy from Proxmox: only guests in the pvewarden pool; updates need tag pvewarden-update,
    reboots need pvewarden-reboot.
  • Updates only from the distribution's official archives (Debian, Ubuntu incl. ESM). Third-party
    repositories and held packages are reported, never installed.
  • A snapshot before every update, and none when nothing is pending. Snapshot descriptions name the
    agent.
  • The user confirms rollback, snapshot deletion and updates without a snapshot in a dialog in the
    MCP client (MCP elicitation, [policy] confirm_destructive).
  • Guests that Proxmox has locked (backup, migration) are left alone. check_updates reports locks
    and running backups.
  • Updates run detached on the guest. A dropped connection or a timeout does not interrupt dpkg;
    update_result picks the result up later.
  • Services are compared before and after an update; newly failed units are reported.
  • Audit log of every call, with the agent's name. Full apt output is kept for every update.
  • pvewarden doctor checks the config, the Proxmox API, the token's privileges (too few and too
    many) and SSH to every guest.

Setup (pvewarden-setup, on the Proxmox host)

  • TUI wizard: pool, minimal role and token, an unprivileged Debian container for the server,
    enrolling guests (LXC and VMs), connection summary.
  • HTTPS certificate signed by the Proxmox cluster CA. Clients fetch the CA from /ca.pem with a
    fingerprint-checked command.
  • Commands: install, enroll, connect, token, renew-cert, upgrade, uninstall;
    --dry-run and an unattended mode (--yes with PVEWARDEN_* variables).
  • The guests' own automatic updates: keep (default), off or on, chosen per enroll.

Tested with Proxmox VE 9.1 and Debian 12/13 and Ubuntu 22.04/24.04/26.04 guests, as containers
and VMs. Claude Code is the MCP client.