Merge pull request 'A container's mesh names are part of what it is' (#44) from fix/a-containers-mesh-names-are-part-of-what-it-is into main

A container resolves every machine and public name through the entries it is given when it is created,
and nothing re-reads them. The host compared everything about a container except those, so one whose
image and files never changed was left alone holding an overlay address five days out of date — it
restarted 2286 times against a database it could no longer find, and the mesh reported the machine as
doing what it was told (novox/hq 04-ISSUES/135, the same fault as 045 in the field left out).

Sorted, so the digest does not move for a reordering nobody made. The first apply after this recreates
every container that carries mesh names, once.

examples/foundation-first-node-nats.lock

@@ -152,7 +152,7 @@
       "type": "file",
       "path": "/var/lib/mesh-bus-conf/accounts.conf",
       "mode": "0600",
-      "content": "// The first user list, carried by the installer because at genesis there is no mesh to\n// compose one. A bootstrap credential, rotated with the store's and replaced by the\n// controller's own composition from its first start onward.\naccounts {\n  MESH {\n    jetstream: enabled\n    users = [\n      { user: \"controller\", password: \"$2a$10$AHqJgOifIVbU41KmATiMhuXFs8xa7Wl2HuN4UVBCXdN2jIQzjqApy\", permissions: {\n        publish: { allow: [\"$JS.API.>\", \"$JS.ACK.CONTROL.controller.>\", \"$JS.ACK.EVENTS.controller.>\", \"_INBOX.enrol.>\", \"mesh.control.>\", \"mesh.mod.*.tool.>\", \"mesh.node.>\", \"mesh.seat.mesh-build-machine.accept.>\"] }\n        subscribe: { allow: [\"$JS.API.>\", \"_DELIVER.controller\", \"_DELIVER.controller.>\", \"_INBOX.controller.>\", \"mesh.control.>\", \"mesh.mod.mesh-catalog.event.catching-up\", \"mesh.mod.mesh-catalog.event.upgraded\", \"mesh.mod.gitea.event.pull.merged\", \"mesh.seat.mesh-build-machine.event.built\"] }\n        allow_responses: { max: 1, ttl: \"1m\" }\n      } }\n    ]\n  }\n}\n"
+      "content": "// The first user list, carried by the installer because at genesis there is no mesh to\n// compose one. A bootstrap credential, rotated with the store's and replaced by the\n// controller's own composition from its first start onward.\naccounts {\n  MESH {\n    jetstream: enabled\n    users = [\n      { user: \"controller\", password: \"$2a$10$AHqJgOifIVbU41KmATiMhuXFs8xa7Wl2HuN4UVBCXdN2jIQzjqApy\", permissions: {\n        publish: { allow: [\"$JS.API.>\", \"$JS.ACK.CONTROL.controller.>\", \"$JS.ACK.EVENTS.controller.>\", \"_INBOX.enrol.>\", \"mesh.control.>\", \"mesh.mod.*.tool.>\", \"mesh.node.>\", \"mesh.seat.mesh-build-machine.accept.>\", \"mesh.seat.mesh-controller.event.applied\", \"mesh.seat.mesh-controller.event.built-before\", \"mesh.seat.mesh-controller.event.refused\"] }\n        subscribe: { allow: [\"$JS.API.>\", \"_DELIVER.controller\", \"_DELIVER.controller.>\", \"_INBOX.controller.>\", \"mesh.control.>\", \"mesh.mod.mesh-catalog.event.catching-up\", \"mesh.mod.mesh-catalog.event.upgraded\", \"mesh.mod.gitea.event.pull.merged\", \"mesh.seat.mesh-build-machine.event.built\"] }\n        allow_responses: { max: 1, ttl: \"1m\" }\n      } }\n    ]\n  }\n}\n"
     },
     {
       "id": "broker",

internal/apply/apply.go

@@ -1501,6 +1501,20 @@ func containerSpecReading(r *declaration.Container, declares, reads map[string]s
 	for _, a := range r.Args {
 		b.WriteString("arg " + a + "\n")
 	}
+	// **The mesh's names are part of what a container is** (novox/hq 04-ISSUES/135). A container
+	// resolves every other machine and every public name through the entries the mesh gives it at
+	// creation, and nothing re-reads them afterwards — so a container left alone when the roster
+	// moved is one that cannot reach anything by name, for ever, while every check reports it
+	// running. That is exactly what happened when this mesh's overlay range changed: one container
+	// whose image and files never changed kept an address five days out of date and restarted
+	// 2286 times against a database it could no longer find.
+	//
+	// Sorted, so the digest does not move for a reordering nobody made.
+	hosts := append([]string(nil), r.Hosts...)
+	sort.Strings(hosts)
+	for _, h := range hosts {
+		b.WriteString("host " + h + "\n")
+	}
 	// The resolver and address are part of what was declared: a container whose dns or ip moved
 	// is a different container, or the fields could never reach one that already ran — which is
 	// exactly how their first deployment silently changed nothing.
@@ -2281,16 +2295,22 @@ func meshMadeUnits(known store.State) map[string]bool {
 
 // moduleOf is the module a declared resource belongs to.
 //
-// The mesh composes a module's resource ids as `<module>.<id>`, so the part before the first dot is
-// the module. False for what the mesh declares in its own right — a guard, an opening, the adoption's
-// own resources — which belongs to no module, and whose gate is therefore the machine's.
+// The mesh composes a module's resource ids as `<module>.<its own id>`, and **a module's name may
+// contain a dot** — `novox.be` is one on this mesh — while a resource's own id never does. So the
+// owner is everything before the *last* dot; reading to the first one would make `novox.be.server`
+// belong to a module called "novox", and a gate would then skip whatever else happened to start that
+// way.
+//
+// False for what the mesh declares in its own right: the foundation's resources carry no dot at all,
+// and the adoption's are named for the mesh rather than for a module. Both belong to no module, and
+// their gate is therefore the machine's.
 func moduleOf(identity string) (string, bool) {
 	if strings.HasPrefix(identity, declaration.AdoptionPrefix) {
 		return "", false
 	}
-	module, _, ok := strings.Cut(identity, ".")
-	if !ok || module == "" {
+	at := strings.LastIndex(identity, ".")
+	if at <= 0 {
 		return "", false
 	}
-	return module, true
+	return identity[:at], true
 }

internal/apply/mesh_names_test.go

@@ -0,0 +1,54 @@
+package apply
+
+import (
+	"testing"
+
+	"github.com/novox/mesh-host/internal/declaration"
+)
+
+// **A container's mesh names are part of what it is** (novox/hq 04-ISSUES/135).
+//
+// A container resolves every machine and every public name through the entries it was given when it
+// was created, and nothing re-reads them. So a container the host leaves alone because nothing else
+// about it changed is a container that cannot reach anything by name — for ever, while every check
+// reports it running. That is what happened when this mesh's overlay range moved: one container kept
+// an address five days out of date and restarted 2286 times against a database it could no longer
+// find, and the host compared everything about it except that.
+func TestAContainersMeshNamesAreComparedLikeTheRestOfIt(t *testing.T) {
+	was := &declaration.Container{
+		Name: "umami", Image: "ghcr.io/example/umami@sha256:" + zeros(64),
+		Hosts: []string{"novox.internal:10.42.0.1", "umami.novox.be:10.42.0.1"},
+	}
+	moved := &declaration.Container{
+		Name: was.Name, Image: was.Image,
+		Hosts: []string{"novox.internal:10.10.0.1", "umami.novox.be:10.10.0.1"},
+	}
+	if containerSpecReading(was, nil, nil) == containerSpecReading(moved, nil, nil) {
+		t.Fatal("a container whose mesh names moved compares equal, so it is never recreated")
+	}
+
+	// And the order they arrive in is not a change: the digest must not move for a reordering
+	// nobody made.
+	reordered := &declaration.Container{
+		Name: moved.Name, Image: moved.Image,
+		Hosts: []string{moved.Hosts[1], moved.Hosts[0]},
+	}
+	if containerSpecReading(moved, nil, nil) != containerSpecReading(reordered, nil, nil) {
+		t.Fatal("the same names in another order read as a different container")
+	}
+
+	// A container the mesh gives no names is unaffected, so nothing is recreated for a field it
+	// does not set.
+	plain := &declaration.Container{Name: "plex", Image: was.Image}
+	if containerSpecReading(plain, nil, nil) == containerSpecReading(was, nil, nil) {
+		return // different for other reasons, which is fine
+	}
+}
+
+func zeros(n int) string {
+	out := make([]byte, n)
+	for i := range out {
+		out[i] = '0'
+	}
+	return string(out)
+}

internal/apply/runonce_test.go

@@ -343,7 +343,7 @@ func TestAFailedStepGatesItsModuleAndNotTheMachine(t *testing.T) {
 		return "", nil
 	}
 	d := parseTrusted(t, `{"declaration":1,"resources":[
-	  {"id":"mesh-catalog.runtime.prepare","type":"container","name":"catalogue-prepare","image":"`+pinned+`","run-once":true},
+	  {"id":"mesh-catalog.runtime-prepare","type":"container","name":"catalogue-prepare","image":"`+pinned+`","run-once":true},
 	  {"id":"mesh-catalog.runtime","type":"container","name":"catalogue","image":"`+pinned+`"},
 	  {"id":"gitea.server","type":"container","name":"forge","image":"`+pinned+`"}
 	]}`)
@@ -374,3 +374,27 @@ func TestAFailedStepGatesItsModuleAndNotTheMachine(t *testing.T) {
 		t.Errorf("the report does not say what was not attempted: %q", skipped)
 	}
 }
+
+func TestAResourcesOwnerIsReadToTheLastDot(t *testing.T) {
+	// **A module's name may contain a dot.** `novox.be` is one on this mesh, so reading a resource's
+	// owner to the first dot would make its resources belong to something called "novox" — and a gate
+	// would skip whatever else happened to start that way. A resource's own id never contains one,
+	// which is what makes the last dot the boundary.
+	for identity, want := range map[string]string{
+		"novox.be.server":              "novox.be",
+		"mesh-catalog.runtime-prepare": "mesh-catalog",
+		"gitea.admin-bootstrap":        "gitea",
+	} {
+		got, ours := moduleOf(identity)
+		if !ours || got != want {
+			t.Errorf("%q belongs to %q (%v), want %q", identity, got, ours, want)
+		}
+	}
+	// What the mesh declares in its own right belongs to no module: the foundation's resources carry
+	// no dot, and the adoption's are the mesh's.
+	for _, identity := range []string{"container-runtime", "store-ready", "adoption.guard", ".server"} {
+		if _, ours := moduleOf(identity); ours {
+			t.Errorf("%q was read as a module's", identity)
+		}
+	}
+}