Repository navigation
docs(range): gandalf is ready for tpl-kali's build (#421, #790) - #1333
Merged
Merged
Conversation
phoenix@pve exists on gandalf with PhoenixBuilder on /vms, range, local and vmbr0, PhoenixIsoCleanup on local and PVEAuditor on the node. Nothing is granted on local-lvm or vmbr1. The Kali 2026.2 installer is on local, checked against the pinned hash, and kali_iso_file defaults to it. No token yet: it is made at the build. Refs #421. Refs #790. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
There was a problem hiding this comment.
🟡 Changes recommended
The runbook unnecessarily grants Gandalf datastore deletion and storage-configuration privileges for a Kali build that generates no answer disc.
2 open issues
What changed in this PR
Prepares gandalf for building the range’s tpl-kali template from phoenix. The template build and its verification remain outstanding under #790.
Changes:
- Sets the Kali ISO default to the installer staged on
gandalf. - Documents applied permissions, separate API credentials, CA trust, and the
rangestorage override. - Records the preparation in the changelog and links the template build into the roadmap.
| File | Description |
|---|---|
packer/variables.pkr.hcl |
Updates the Kali ISO default and explanatory comment. |
docs/runbooks/build-the-lab-templates.md |
Documents gandalf permissions and build setup. |
docs/roadmap.md |
Links the attack VM’s template-build dependency. |
docs/changelog.md |
Records completed host preparation and remaining verification. |
Review effort: Balanced
Gerrrt
enabled auto-merge (squash)
November 9, 2026 04:37
Kali's build uploads no answer disc, so Datastore.Allocate on gandalf's local bought nothing. The grant and the role are removed on gandalf. Refs #790. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Gerrrt
added a commit
that referenced
this pull request
Nov 9, 2026
Resolves the conflicts with #1333, which recorded gandalf's half of templates §5 step 2. Its text and table are kept, with this branch's build results added. Where the two disagree on what gandalf holds, the docs follow the token's live privileges: PhoenixBuilder on vmbr1 (used by make-attack-vm.sh) and PhoenixIsoCleanup on local are both granted, and the new changelog entry names the one it corrects. Refs #421 Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.


What changed
gandalfis ready fortpl-kali's build (build-the-lab-templates.md§5, step 2'sgandalfhalf).phoenix@pvenow exists ongandalf, withPhoenixBuilderon/vms,/storage/range,/storage/localandvmbr0, andPVEAuditoron/nodes/gandalf. APhoenixIsoCleanupgrant onlocalwas revoked after review, because Kali's build uploads no answer disc. The Kali 2026.2 installer is ongandalf'slocal, andkali_iso_filenow defaults to it. The runbook records the grants and howphoenixshould holdgandalf's token. The roadmap and changelog are updated.Why
Refs #421 and #790. The roadmap names §5, the attack VM, as next. It clones
tpl-kali, which Packer builds fromphoenixthroughgandalf's own API. That API had no user, no role and no ISO.Blast radius
gandalfonly. It has no guests. The new user has no token yet, so nothing can authenticate as it. Its grants leave outlocal-lvm(ADR-0096 keeps range disks off the M.2) andvmbr1(the range NIC belongs to the clone). They grant nothing under/nodesbeyond audit, so the token cannot touch the host firewall. The role leaves outSaruman'sPool.*.secrets/*.sops.yaml. No token was made. The runbook has the operator make it at the build, so its secret goes straight tophoenix.Verification
Ran on
gandalf, as root:The API certificate: issuer
PVE Cluster Manager CA, SANs include10.0.30.30, no custompveproxy-ssl.pem. Nothing was reissued.pveum acl list | grep phoenixshows the five grants above.pveum role listshowsPhoenixBuilderwith the jumpbox list plusVM.GuestAgent.Audit, and noPhoenixIsoCleanup.pvesh get /nodesreadsgandalf.The ISO was downloaded from cdimage.kali.org and passed
sha256sum -cagainst the hash pinned inscripts/collect-iso-store-state.sh.pvesm list local --content isolistslocal:iso/kali-linux-2026.2-installer-amd64.iso.Not run: the build, the smoke test and the
-forcerebuild. Those are Build tpl-kali (VMID 902) on ifrit once it exists #790's, fromphoenix.scripts/check_docs.pypasses.make validateexits 0 with 43 checks skipped, because this host has no docker daemon and no pipx. That includes packer validate, shellcheck and markdownlint.make validatepasses (it passes with 43 skipped; see above)Deployed to the lab and confirmed working (
gandalf's user, grants and ISO, read back)Docs updated (
docs/, service README, ordocs/roadmap.md)🤖 Generated with Claude Code