Keep what each machine says of what it runs, raise it, and gate on it (hq ADR 0240, to-be 48 Phase A)

The gate judged a module by what the mesh saw from outside, so a container that
crash-looped after it applied passed it. Each machine's node-engine now states
the health of every long-running resource it runs; the controller keeps the
newest statement per machine, raises module.<module>.<machine>.unhealthy on the
second statement in a row, clears it on the first that does not say it, and the
gate passes a module only when every long-running resource of it is stated
healthy since the send. An engine that states nothing is judged as before.
This commit is contained in:
jochen
2026-10-07 02:28:16 +02:00
parent 5d3e52219b
commit 1cc6a2d759
22 changed files with 997 additions and 5 deletions
+3 -1
View File
@@ -86,7 +86,9 @@ var WritersTable = []WriterRow{
Others: "read", Subjects: []string{"mesh.node.*.declare"}, Writes: isController},
{State: "a machine's applied state and its report", Writer: "the node-engine's apply queue",
KeptIn: "the machine; the report on the bus", Others: "the reconcile and a delivery enqueue, never apply",
Subjects: []string{"mesh.control.*.report"}, Writes: ownMachine},
// And its health statement between reports (novox/hq ADR 0240): the same writer stating the same
// machine, inside the grant it already had (`mesh.control.<its own>.>`).
Subjects: []string{"mesh.control.*.report", "mesh.control.*.health"}, Writes: ownMachine},
{State: "the controller lease", Writer: "the controller instance holding it", KeptIn: "key-value " + LeaseBucket,
Others: "a candidate waits", Subjects: kvOf(LeaseBucket), Writes: isController},
{State: "plans and their tiers", Writer: "controller (lease holder), compare-and-set on the plan's revision",