Gerrrt · Nov 9, 2026
diff --git a/docs/architecture.md b/docs/architecture.md
@@ -186,7 +186,7 @@ hole from the monitoring VLAN into the monitored one.
 | --- | --- | --- | --- |
 | `prometheus` (10.0.99.20) | 🔴 99 | [`stacks/observability`](../stacks/observability) | Prometheus, Alertmanager, Loki, Grafana, snmp-exporter, blackbox-exporter, speedtest-tracker, docker-socket-proxy, Alloy |
 | `Saruman` (10.0.30.20) | 🟢 30 | *(none — and none intended)* | Proxmox VE 9, hosting twenty-one guests: `alexander`, built 2026-09-05 ([#262](https://github.com/Gerrrt/HomeLab/issues/262)); `phoenix`, built 2026-09-20 ([#436](https://github.com/Gerrrt/HomeLab/issues/436)); `odin`, built 2026-09-27 for the security tooling (ADR-0030); `fenrir`, built 2026-09-30 as the Zeek sensor ([#437](https://github.com/Gerrrt/HomeLab/issues/437)); `golem`, built 2026-10-03 as the backup server ([#485](https://github.com/Gerrrt/HomeLab/issues/485)); `eden`, built 2026-10-07 as the BloodHound guest ([#451](https://github.com/Gerrrt/HomeLab/issues/451)), off between sessions; `garuda`, built 2026-10-10 as the analyst workstation ([#921](https://github.com/Gerrrt/HomeLab/issues/921)); and ADR-0029's six-machine domain, built by hand 2026-09-24 to 2026-09-25 ([#414](https://github.com/Gerrrt/HomeLab/issues/414)), of which `carbuncle` and `siren` run per session; and all eight of [ADR-0090](adr/0090-test-the-dotfiles-os-layers-on-on-demand-saruman-guests.md)'s on-demand dotfiles test VMs, `dot-debian`, `dot-fedora` and `dot-windows`, built 2026-10-08, and `dot-opensuse`, `dot-arch`, `dot-alpine`, `dot-gentoo` and `dot-nixos`, built 2026-10-09 ([#920](https://github.com/Gerrrt/HomeLab/issues/920)). Alloy agent (native package). It runs no compose stack by decision, not by omission: Docker would rewrite the iptables its own firewall relies on ([ADR-0014](adr/0014-put-ifrit-on-imaginationlan-and-give-the-targets-no-route.md)), which is why the agent here is the native package and why `stacks/lab` runs in a guest |
-| `gandalf` (10.0.30.30) | 🟢 30 | *(none — and none intended)* | Proxmox VE 9, the range host, installed 2026-11-07 ([#421](https://github.com/Gerrrt/HomeLab/issues/421)). No guests yet. Guest disks go on the ZFS pool `range`, targets go on the portless `vmbr1`, and it is off between sessions ([ADR-0017](adr/0017-buy-ifrit-for-iops-and-keep-the-range-disposable.md), [ADR-0106](adr/0106-name-the-range-host-gandalf-and-leave-ifrit-to-its-disabled-amt.md)) |
+| `gandalf` (10.0.30.30) | 🟢 30 | *(none — and none intended)* | Proxmox VE 9, the range host, installed 2026-11-07 ([#421](https://github.com/Gerrrt/HomeLab/issues/421)). One guest, the attack VM `ragnarok` (VMID 130), made 2026-11-09 by [`scripts/make-attack-vm.sh`](../scripts/make-attack-vm.sh) from template 902 and dual-homed: a DHCP lease on VLAN 30, and `172.30.30.10` on `vmbr1`. No targets yet. Guest disks go on the ZFS pool `range`, targets go on the portless `vmbr1`, and it is off between sessions ([ADR-0017](adr/0017-buy-ifrit-for-iops-and-keep-the-range-disposable.md), [ADR-0106](adr/0106-name-the-range-host-gandalf-and-leave-ifrit-to-its-disabled-amt.md)) |
 | `alexander` (10.0.30.40) | 🟢 30 | [`stacks/lab`](../stacks/lab) | Prometheus, Loki, Grafana, Alloy and its Docker socket proxy, and Caddy as the ingest proxy in front of the first two ([#834](https://github.com/Gerrrt/HomeLab/issues/834)) — the lab's own observability, which never remote-writes to VLAN 99 ([ADR-0007](adr/0007-defensive-estate-and-offensive-range.md), [ADR-0020](adr/0020-run-the-lab-stack-in-a-guest-with-its-own-prometheus.md)). A guest on `Saruman`, not the hypervisor; Alloy agent (Docker) |
 | `odin` (10.0.30.60) | 🟢 30 | [`stacks/soc`](../stacks/soc) | Wazuh (indexer, manager, dashboard), Velociraptor and Alloy: the security half of ADR-0007, placed by [ADR-0030](adr/0030-give-the-security-tooling-its-own-guest-and-its-own-stack.md) on a second guest on `Saruman` because `alexander`'s 8 GiB cannot hold both. Built 2026-09-27 by [`build-the-soc-guest.md`](runbooks/build-the-soc-guest.md), the stack authored and CI-validated ahead of the guest the way `stacks/lab` was ahead of `alexander` ([#266](https://github.com/Gerrrt/HomeLab/issues/266), [#267](https://github.com/Gerrrt/HomeLab/issues/267)); the six domain machines enrol as agents by GPO, which is the day those two issues close. Its Alloy pushes to `alexander`, never to VLAN 99, and the indexer's health is the one series that crosses into the lab's Prometheus. Alloy agent (Docker) |
 | `fenrir` (10.0.30.90) | 🟢 30 | [`stacks/sensor`](../stacks/sensor) | Zeek on a `tc` mirror of `Saruman`'s lab bridge, so that the domain's east-west traffic, which crosses no router and so never reaches Suricata on `morpheus`, is observed at all ([#437](https://github.com/Gerrrt/HomeLab/issues/437), [ADR-0068](adr/0068-mirror-the-lab-bridge-to-zeek-with-tc-not-open-vswitch.md)). A guest on `Saruman`, VMID 190, with a second NIC alone on the port-less `vmbr1` that receives only the mirror's copies. Its logs go to `alexander`, never to VLAN 99; whether the mirror is delivering is answered on the hypervisor by `homelab_zeek_mirror_active`. Built 2026-09-30 by [`build-the-sensor-guest.md`](runbooks/build-the-sensor-guest.md), and proved by rebooting `Saruman` with the mirror's timer disabled: the gauge fell to 0 and `ZeekMirrorInactive` fired. Alloy agent (Docker) |

diff --git a/docs/changelog.md b/docs/changelog.md
@@ -19,6 +19,39 @@ docstring gives: it is a record, not a claim about now.
 
 ## 2026-11-09
 
+- **[#421](https://github.com/Gerrrt/HomeLab/issues/421),
+  [#790](https://github.com/Gerrrt/HomeLab/issues/790): `tpl-kali` (902) is
+  built on `gandalf`, and the attack VM `ragnarok` is up,
+  `build-the-playground.md` §5.**
+  - **Access.** `gandalf` is not clustered, so it got its own `phoenix@pve`
+    user, `PhoenixBuilder` and `PhoenixIsoCleanup` roles and
+    `phoenix@pve!builder` token. The role is `build-the-jumpbox.md` §4's plus
+    `VM.GuestAgent.Audit`. Its ACLs cover `range`, `local`, and both
+    `vmbr0` and `vmbr1`. The two builds, the two smoke tests and the attack
+    VM asked for nothing more. `phoenix` trusts `gandalf`'s own cluster CA
+    and keeps the token in `gandalf.env`, apart from `phoenix.env`.
+  - **The builds.** The ISO is the 2026.2 installer already pinned for 910,
+    fetched onto `gandalf`'s `local` and checked against that hash. The
+    first build stopped at validation, because the variables file refuses
+    an empty `PKR_VAR_build_password` even for Kali. Sourcing `phoenix.env`
+    under `gandalf.env` fixed it, and §5 says so now. The build took
+    10 min 33 s and the `-force` rebuild 10 min 41 s. `packer-smoke.sh 902`
+    passed after each, which is #790's acceptance. 910's boot-order and 20 s
+    fixes were not needed here. `ufw` is inactive on `phoenix`, so no port
+    was opened.
+  - **The attack VM.** `ragnarok`, VMID 130, is a full clone on `range` with
+    4 cores and 8 GiB, made by the new `scripts/make-attack-vm.sh`. `net0` is
+    on `vmbr0` by DHCP and took `10.0.30.111`. `net1` is on `vmbr1` at
+    `172.30.30.10/24` with no gateway. Both have the firewall flag off. The
+    script wrote `99-no-forward.conf` and ran §8's third check: both
+    sysctls `0`, no NAT in nft or iptables, and one default route, via
+    `10.0.30.1`.
+  - **#1196's check.** On the fresh clone it read `running`, 0 failed units
+    and 0 accounts expiring at the epoch, the check
+    [#1196](https://github.com/Gerrrt/HomeLab/issues/1196) owed 902.
+  - **What is not done.** Sliver is not installed, there are no targets
+    (§6), and §8's first, second and fourth checks and §9 have not run.
+
 - **[#421](https://github.com/Gerrrt/HomeLab/issues/421): `gandalf`'s
   management plane is closed, `build-the-playground.md` §4.** The firewall
   had been off since the install, so ADR-0014's rules did not exist yet and

diff --git a/docs/diagrams/current/network.png b/docs/diagrams/current/network.png
diff --git a/docs/diagrams/current/network.svg b/docs/diagrams/current/network.svg
diff --git a/docs/network.md b/docs/network.md
@@ -713,7 +713,8 @@ Where things get broken on purpose.
   away before the attack VM starts taking leases from the same pool.
 - A second hypervisor, `gandalf`, carries the attack tooling and the
   deliberately-vulnerable targets. It is installed (Proxmox VE 9,
-  2026-11-07) and has no guests yet. It sits on this segment single-homed, on
+  2026-11-07) and has one guest, the attack VM `ragnarok` (VMID 130, made
+  2026-11-09). It sits on this segment single-homed, on
   `neo`'s untagged `ether20`, at `10.0.30.30`: a static below `.100`, with a
   reservation. Its Intel AMT, which would be the name `ifrit`, is disabled
   ([ADR-0106](adr/0106-name-the-range-host-gandalf-and-leave-ifrit-to-its-disabled-amt.md)).
@@ -726,7 +727,7 @@ Where things get broken on purpose.
   is the third private block in the house and the only one that is not routed
   anywhere, which is what makes a `172.30.30.x` source in a firewall block a
   leak reporting itself. The attack VM takes a lease from the pool like any
-  other guest. The build is
+  other guest, with no reservation, and holds `172.30.30.10` on the bridge. The build is
   [`build-the-playground.md`](runbooks/build-the-playground.md).
 
 - **The WireGuard peers live on `172.31.0.0/24`, and it is routed rather than

diff --git a/docs/roadmap.md b/docs/roadmap.md
@@ -240,8 +240,8 @@ Closes on BloodHound running where nothing attacks it.
   [ADR-0074](adr/0074-build-the-lab-templates-with-packer-from-phoenix.md):
   901 Ubuntu, 911 Windows 11 and 912 Server 2025 built from `smaug-iso`,
   each clone passing `scripts/packer-smoke.sh`, SSH included,
-  → [runbook](runbooks/build-the-lab-templates.md); Kali's
-  template waits for its build on `gandalf`, [#790](https://github.com/Gerrrt/HomeLab/issues/790)) →
+  → [runbook](runbooks/build-the-lab-templates.md); Kali's 902
+  built on `gandalf` 2026-11-09, [#790](https://github.com/Gerrrt/HomeLab/issues/790)) →
   [#445](https://github.com/Gerrrt/HomeLab/issues/445) OpenTofu (done
   2026-10-03,
   [ADR-0076](adr/0076-provision-lab-guests-with-opentofu-and-encrypt-its-state-from-the-first-apply.md):
@@ -286,9 +286,11 @@ Gated on the domain, on a household observation, or on something to publish.
 
 - **[#421](https://github.com/Gerrrt/HomeLab/issues/421) Build the range on
   `gandalf`.** Gated on #414 and the SOC: an attack VM pointed at an
-  uninstrumented estate teaches nothing. Both are met. The runbook's §1–§4
-  are behind it: the host, its bridges, `Saruman` at `.20` and the closed
-  management plane. Next is §5, the attack VM, then §6–§9, in order.
+  uninstrumented estate teaches nothing. Both are met. The runbook's §1–§5
+  are behind it: the host, its bridges, `Saruman` at `.20`, the closed
+  management plane, and the attack VM `ragnarok` on template 902. Next is §6,
+  the targets, then §7–§9, in order. Sliver goes on `ragnarok` before the
+  first exercise.
   The name is
   [ADR-0106](adr/0106-name-the-range-host-gandalf-and-leave-ifrit-to-its-disabled-amt.md)'s.
   → [runbook](runbooks/build-the-playground.md)

diff --git a/docs/runbooks/build-the-analyst-workstation.md b/docs/runbooks/build-the-analyst-workstation.md
@@ -553,6 +553,9 @@ the console password.
   - **Still owed:** the same check on 902 (gandalf's Kali, same preseed) when it
     is next built — [#1196](https://github.com/Gerrrt/HomeLab/issues/1196),
     tied to [#790](https://github.com/Gerrrt/HomeLab/issues/790).
+  - **Paid 2026-11-09:** 902 was built twice on `gandalf`, and its clone
+    `ragnarok` read `running`, 0 failed units and 0 accounts at the epoch
+    ([`build-the-lab-templates.md`](build-the-lab-templates.md) §5).
 - **§12, phase 2, 2026-10-10**
   ([#1163](https://github.com/Gerrrt/HomeLab/pull/1163),
   [ADR-0092](../adr/0092-enrol-garuda-in-odins-soc-with-debian-packages-staged-on-odin.md)).

diff --git a/docs/runbooks/build-the-lab-templates.md b/docs/runbooks/build-the-lab-templates.md
@@ -1,8 +1,8 @@
 # Runbook: Build the lab's VM templates with Packer, from `phoenix`
 
 **Target:** templates 901, 903–912 on `Saruman`, built from
-`phoenix` through the Proxmox API. 902, Kali, is written here and built on
-`gandalf`, the range host, installed 2026-11-07 and not yet built on. 910 is the same Kali build on `Saruman`, for
+`phoenix` through the Proxmox API. 902, Kali, is built on `gandalf`, the
+range host, through `gandalf`'s own API (§5, first built 2026-11-09). 910 is the same Kali build on `Saruman`, for
 `garuda` (§5b, [ADR-0091](../adr/0091-put-a-kali-purple-analyst-workstation-on-saruman.md)). 903 (Debian), 904 (Fedora), 905 (openSUSE
 Tumbleweed), 906 (Arch), 907 (Alpine), 908 (Gentoo) and 909 (NixOS) are the
 dotfiles OS layers' templates
@@ -488,34 +488,93 @@ node. The preseed is served over Packer's HTTP server on port 8800 of
 `phoenix`, because Debian's installer reads one from a URL or its own medium,
 not from a second disc.
 
-1. Admit the guest to that port on `phoenix` for the length of the build. If
-   `ufw` is active, run `sudo ufw allow from 10.0.30.0/24 to any port 8800 proto tcp`,
-   and delete the rule afterwards. Debian's build (§4) needs this step and
-   no other here.
-2. Grant `PhoenixBuilder` on `gandalf`'s storage and bridge, and trust
-   `gandalf`'s CA on `phoenix`, as §2 did for `Saruman`. Unless `gandalf` joins
-   `Saruman` in a cluster, it is its own API. Point `PROXMOX_URL` at it for
-   this build, and admit `phoenix` to its 8006 the way ADR-0043 did on
-   `Saruman`.
-3. Build, naming the ISO actually on `gandalf` and **its** storage.
-   `disk_storage` defaults to `large_data`, which is `Saruman`'s pool, and
-   the disk creation fails without the override:
+> **Ran 2026-11-09** ([#790](https://github.com/Gerrrt/HomeLab/issues/790),
+> [#421](https://github.com/Gerrrt/HomeLab/issues/421)). Steps 2–5 below are
+> what ran. The first build took 10 min 33 s and the `-force` rebuild
+> 10 min 41 s. `packer-smoke.sh 902` passed after each. `gandalf`'s 10 s
+> `boot_wait` was enough with no boot order, unlike 910 on `Saruman`. The
+> clone `ragnarok` then read `running` with 0 failed units and no account
+> expiring at the epoch, the check
+> [#1196](https://github.com/Gerrrt/HomeLab/issues/1196) owed 902.
+
+1. **Port 8800 on `phoenix`.** If `ufw` is active, run
+   `sudo ufw allow from 10.0.30.0/24 to any port 8800 proto tcp` for the build
+   and delete the rule afterwards. On 2026-11-09 `ufw` was inactive on
+   `phoenix`, so there was nothing to open. Debian's build (§4) needs this
+   step and no other here.
+2. **The user, the roles and the token, on `gandalf`, as root.** `gandalf` is
+   not clustered with `Saruman`
+   ([ADR-0039](../adr/0039-decline-proxmox-clustering-while-ifrit-is-the-range.md)),
+   so it is its own API with its own users. The role is
+   [`build-the-jumpbox.md`](build-the-jumpbox.md) §4's plus §2's
+   `VM.GuestAgent.Audit`. The storage is `range`, and the bridges are both
+   `vmbr0` and `vmbr1`, because the attack VM's second NIC is on the range
+   bridge:
 
    ```bash
-   packer build -only='kali.proxmox-iso.kali' \
-     -var kali_iso_file=local:iso/<the iso> \
-     -var disk_storage=<gandalf's guest storage> \
-     packer/
+   pveum role add PhoenixBuilder --privs "VM.Allocate VM.Clone VM.Config.CDROM VM.Config.CPU VM.Config.Cloudinit VM.Config.Disk VM.Config.HWType VM.Config.Memory VM.Config.Network VM.Config.Options VM.Console VM.PowerMgmt VM.Audit VM.Snapshot VM.GuestAgent.Audit Datastore.AllocateSpace Datastore.AllocateTemplate Datastore.Audit SDN.Use"
+   pveum role add PhoenixIsoCleanup --privs "Datastore.Allocate"
+   pveum user add phoenix@pve --comment "deployment host, ADR-0043"
+   pveum acl modify /vms --users phoenix@pve --roles PhoenixBuilder
+   pveum acl modify /storage/local --users phoenix@pve --roles PhoenixBuilder,PhoenixIsoCleanup
+   pveum acl modify /storage/range --users phoenix@pve --roles PhoenixBuilder
+   pveum acl modify /sdn/zones/localnetwork/vmbr0 --users phoenix@pve --roles PhoenixBuilder
+   pveum acl modify /sdn/zones/localnetwork/vmbr1 --users phoenix@pve --roles PhoenixBuilder
+   pveum acl modify /nodes/gandalf --users phoenix@pve --roles PVEAuditor
+   pveum acl list | grep phoenix
+   pveum user token add phoenix@pve builder --privsep 0
    ```
 
-4. Prove it on `gandalf`, as §6 does on `Saruman`:
+   Paste this into the node's **Shell** in the web UI, or into an SSH session
+   from Hicks. On 2026-11-09 a first paste went into a zsh on `phoenix`,
+   where `pveum` does not exist. The paste handling there also turned every
+   space after the first line into a literal `\s`. Nothing ran, and nothing
+   needed undoing. No privilege beyond these was asked for by the two
+   builds, the two smoke tests or `scripts/make-attack-vm.sh`.
+
+   **The installer goes on `gandalf`'s own `local`.** It is the same 2026.2
+   ISO as 910's, checked against the hash
+   [`collect-iso-store-state.sh`](../../scripts/collect-iso-store-state.sh)
+   pins:
+
+   ```bash
+   cd /var/lib/vz/template/iso && wget -nc https://cdimage.kali.org/kali-2026.2/kali-linux-2026.2-installer-amd64.iso && echo "6dbefacc95e3b556c19c48e8bae39b8b505e2d3a1aba0bfb7ab62b036c3d2ba3  kali-linux-2026.2-installer-amd64.iso" | sha256sum -c -
+   ```
+
+3. **`phoenix` trusts `gandalf`'s CA and holds its token apart.** `gandalf`
+   serves its own `PVE Cluster Manager CA` certificate, naming `10.0.30.30`
+   (read 2026-11-09, so it needs no fix of the kind §2 records for
+   `Saruman`). Relay the CA from Hicks:
+   `ssh gandalf cat /etc/pve/pve-root-ca.pem | ssh phoenix 'cat > ~/gandalf-pve-root-ca.crt'`.
+   Install it on `phoenix` as
+   `/usr/local/share/ca-certificates/gandalf-pve-root-ca.crt`, as §2 does
+   for `Saruman`'s. The token goes in **`~/.config/proxmox/gandalf.env`**,
+   mode 600, with its own `PROXMOX_URL=https://10.0.30.30:8006/api2/json`,
+   `PROXMOX_TOKEN_ID=phoenix@pve!builder` and `PROXMOX_TOKEN_SECRET`. It is
+   not added to `phoenix.env`, so a build for `Saruman` can never be pointed
+   at `gandalf` by accident, or the reverse. Prove it with §2's `curl …/version`
+   line, with `gandalf.env` sourced in place of `phoenix.env`.
+4. **Build.** Source `phoenix.env` first and `gandalf.env` over it. The
+   variables file refuses an empty `PKR_VAR_build_password` for every source,
+   Kali included, and that password lives in `phoenix.env`. The `PROXMOX_*`
+   lines in `gandalf.env` then win. `disk_storage` defaults to `large_data`,
+   which is `Saruman`'s pool, so name `range`:
+
+   ```bash
+   set -a; . ~/.config/proxmox/phoenix.env; . ~/.config/proxmox/gandalf.env; set +a
+   packer init packer/
+   packer build -force -only='kali.proxmox-iso.kali' -var disk_storage=range packer/
+   ```
+
+   `kali_iso_file` defaults to the ISO from step 2.
+5. **Prove it**, with `gandalf.env` sourced:
 
    ```bash
    PROXMOX_NODE=gandalf scripts/packer-smoke.sh 902
    ```
 
-   Then run §8's second `-force` build and smoke test, which is #790's
-   acceptance.
+   Then run the build and the smoke test a second time, which is #790's
+   acceptance and §8's rule.
 
 ## 5b. Kali on `Saruman`, for `garuda`