The beds raise a mesh through an installer that carries a builder #22

2026-09-13T09:17:23Z

jschoubben commented

2026-09-13 09:17:23 +00:00

Both genesis beds now pass the installer a repository and a commit, and refuse to run without them rather than raising a machine the installer cannot finish. The installer they build carries the builder rather than the control plane.

The bed checks the control plane was built, not carried. The pivot checks proved the running control plane is pinned to a digest this mesh's registry serves — which a carried image satisfies just as well. What the installer now exists to make true is that a build happened, from the commit the bed asked for, and that was being printed and not checked.

The bed resolves an artifact the way the builder would. It pre-builds these images on the workstation and stocks them, which is the lab standing in for the builder — so it must do what the builder does. Without this the unresolved field travels to the machine, whose declaration language has no such field, and the whole declaration is refused.

The image scripts stage the toolkit from its own checkout, not from a symlink that may not be one. They copied the runtime's installed tree and relied on the toolkit inside it being a link into the sibling repository — true only where somebody had linked them by hand, and false as soon as dependencies are installed the ordinary way. The image still built, and every entry point in it pointed at nothing.

Single-machine bed passes. On four machines genesis, the joins and the overlay across the access point all pass; the suite's overall verdict is still red, for two third-party images whose downloads did not complete and a firewall service that is not running on that image — none of which this branch touches.

Both genesis beds now pass the installer a repository and a commit, and refuse to run without them rather than raising a machine the installer cannot finish. The installer they build carries the builder rather than the control plane. **The bed checks the control plane was built, not carried.** The pivot checks proved the running control plane is pinned to a digest this mesh's registry serves — which a carried image satisfies just as well. What the installer now exists to make true is that a build happened, from the commit the bed asked for, and that was being printed and not checked. **The bed resolves an artifact the way the builder would.** It pre-builds these images on the workstation and stocks them, which is the lab standing in for the builder — so it must do what the builder does. Without this the unresolved field travels to the machine, whose declaration language has no such field, and the whole declaration is refused. **The image scripts stage the toolkit from its own checkout**, not from a symlink that may not be one. They copied the runtime's installed tree and relied on the toolkit inside it being a link into the sibling repository — true only where somebody had linked them by hand, and false as soon as dependencies are installed the ordinary way. The image still built, and every entry point in it pointed at nothing. Single-machine bed passes. On four machines genesis, the joins and the overlay across the access point all pass; the suite's overall verdict is still red, for two third-party images whose downloads did not complete and a firewall service that is not running on that image — none of which this branch touches.

jschoubben added 4 commits 2026-09-13 09:17:23 +00:00

Stage the sdk from its own checkout, not from a symlink that may not be one 1ba237a634

Both image scripts copied the runtime's installed tree and relied on the sdk
inside it being a link into the sibling repository. That is true only where
somebody linked them by hand, and false as soon as the dependencies are
installed the ordinary way — which fetches the sdk as sources with nothing
compiled. The image still built, and every entry point in it pointed at
nothing.

The lab's installer carries a builder, and genesis is told what to build 607ea241c7

Both beds now pass a repository and a commit, and refuse to run without them
rather than raising a machine the installer cannot finish.

The bed checks the control plane was built, not carried fb18807000

The pivot checks proved the running control plane is pinned to a digest this
mesh's registry serves, which a carried image satisfies just as well. What the
installer now exists to make true is that a build happened, from the commit the
bed asked for — and that was printed and not checked.

The bed resolves an artifact the way the builder would d27f24cf3e

It pre-builds these images and stocks them, which is the lab standing in for the
builder — so it must do what the builder does and replace the artifact with the
reference the machine holds. Without it the unresolved field travels to the
machine and the whole declaration is refused.

jschoubben merged commit 36a5ad758b into main

2026-09-13 09:17:29 +00:00

Sign in to join this conversation.

1 Participants

Notifications

Due Date

No due date set.

Dependencies

No dependencies set.

Reference: novox/mesh-lab#22