diff --git a/04-ISSUES/169-a-machine-shares-its-files-and-the-mesh-does-not-know/00-report.md b/04-ISSUES/169-a-machine-shares-its-files-and-the-mesh-does-not-know/00-report.md index f15141e..f51111c 100644 --- a/04-ISSUES/169-a-machine-shares-its-files-and-the-mesh-does-not-know/00-report.md +++ b/04-ISSUES/169-a-machine-shares-its-files-and-the-mesh-does-not-know/00-report.md @@ -39,20 +39,32 @@ Under the mesh as it stands, this arrangement has no expression and one failure ## The proposal (the operator's, 2026-09-30) -A **file-sharing module** — NFS first, Samba the same shape — that: +**File sharing is a core seat — a role each machine has, defined by the control plane — and, as +with package registries (ADR 0109), one seat per protocol, so several modules occupy the family:** -- declares the exported paths as `accesses` (ADR 0051: it owns nothing about them, never creates, +| seat | scope | delivers | held by | +|---|---|---|---| +| `node-nfs-share` | node | `nfs-share` | an `nfs` module | +| `node-smb-share` | node | `smb-share` | a `samba` module | +| … (`node-webdav-share`) | node | … | whatever comes next | + +Named for their scope (ADR 0121), one holder per node (ADR 0110), each carrying its protocol +(ADR 0129). A machine may hold both — nfs and samba on ace — and one module may hold several +(0109: "gitea may hold several seats at once"). Adding a protocol is adding a seat and a +provision, not widening one. + +The holder module: + +- declares the exported paths as `accesses` (ADR 0051: it owns nothing about them — never creates, chowns or removes), and *which* paths as the assignment's settings (ADR 0046/0112); - writes the share configuration (`/etc/exports`, `smb.conf`) as mesh-managed files and drives the units, like `dnsmasq`/`sshd` do for theirs; -- declares its endpoints (`nfs` 2049/tcp, `smb` 445/tcp, …) so the reach — internal, or the LAN +- declares its endpoints (`nfs` 2049/tcp; `smb` 445/tcp, …) so the reach — internal, or the LAN once 154 has an answer — is the assignment's, and converge keeps them open; -- **defines and holds a node-scoped seat** (`file-share`, one holder per machine, as ADR 0126 - lets a module do) — the machine's one answer for "where are the files"; -- **provides** `file-share` at mesh scope, serving the export path(s) and protocol, so a consumer - on another node `requires` it and reads `${bound:file-share:at}` and the path from its binding - instead of a hand-typed mount; a pair credential where the protocol has one (Samba user), none - for `sec=sys` NFS. +- **provides** the seat's provision, serving the export path(s), so a consumer on another node + `requires nfs-share` (or `smb-share`) and reads `${bound:nfs-share:at}` and the path from its + binding instead of a hand-typed mount — a pair credential where the protocol has one (a Samba + user), none for `sec=sys` NFS. What it would settle: ace's library becomes reachable from the mesh by declaration, the two host services get an owner, converge stops being a trap for them, and a media module on another machine @@ -60,10 +72,10 @@ services get an owner, converge stops being a trap for them, and a media module ## Open questions for the decision -- The seat's name and whether Samba and NFS are one seat with two protocols (ADR 0129: a seat - carries the protocol of its role) or two seats. - Whether an NFS export over the overlay is an `internal` reach of the same endpoint or a second export line — NFS authorises by client address, so the mesh range and the LAN range are two entries in one file. - How a consumer's binding expresses a *path* to mount (today bindings carry `at`, `port`, `as` and - whatever the provider `serves`). + whatever the provider `serves`), and whether one share can serve several paths. +- Whether the seat should exist before its first module (a system seat is a decision, ADR 0110) — + the decision that adds the two seats can be the one that accepts this proposal.