novox/mesh-controller
6
0
Fork 0
Code Issues Pull Requests 1 Releases Wiki Activity

Merge pull request 'Raise a machine's network from what its engine says, once (hq ADR 0241)' (#113) from feat/machine-network-health into main

Browse Source
This commit was merged in pull request #113.
This commit is contained in:
mesh-admin
2026-10-07 18:16:10 +00:00
parent f6aea4bfbb b8bbf9c79f
commit 85d664438f
10 changed files with 966 additions and 9 deletions
Download Patch File Download Diff File Expand all files Collapse all files
internal/inventory/health.go
+42 -7
View File
@@ -40,6 +40,29 @@ type NodeHealth struct {
Resources []ResourceHealth
// Streaks is, per module, how many statements in a row said a resource of it was unhealthy.
Streaks map[string]int
// Network is the machine's own networking as its engine said it (novox/hq ADR 0241); nil from an
// engine older than that judging.
Network *NetworkHealth
}
// NetworkHealth is a machine's networking as its engine said it (ADR 0241).
type NetworkHealth struct {
State string `json:"state"`
Since time.Time `json:"since"`
Parts []NetworkPart `json:"parts"`
}
// NetworkPart is one part of it: resolv-conf, names, tunnel, bus or route.
type NetworkPart struct {
Part string `json:"part"`
State string `json:"state"`
Reason string `json:"reason,omitempty"`
Said string `json:"said,omitempty"`
Writer string `json:"writer,omitempty"`
Owner string `json:"owner,omitempty"`
Toward []string `json:"toward,omitempty"`
Since time.Time `json:"since"`
Streak int `json:"streak,omitempty"`
}
// HealthOf is a machine's newest statement; false when its node-engine has never stated one.
@@ -60,7 +83,7 @@ func (i *Inventory) Healths(ctx context.Context) (map[string]NodeHealth, error)
func (i *Inventory) healths(ctx context.Context, only string) (map[string]NodeHealth, error) {
rows, err := i.store.Pool().Query(ctx,
`select n.name, h.contract, h.said_at, h.heard_at, h.resources, h.streaks
`select n.name, h.contract, h.said_at, h.heard_at, h.resources, h.streaks, h.network
from node_health h join node n on n.id = h.node
where $1 = '' or n.name = $1`, only)
if err != nil {
@@ -70,8 +93,8 @@ func (i *Inventory) healths(ctx context.Context, only string) (map[string]NodeHe
out := map[string]NodeHealth{}
for rows.Next() {
var h NodeHealth
var resources, streaks []byte
if err := rows.Scan(&h.Node, &h.Contract, &h.SaidAt, &h.HeardAt, &resources, &streaks); err != nil {
var resources, streaks, network []byte
if err := rows.Scan(&h.Node, &h.Contract, &h.SaidAt, &h.HeardAt, &resources, &streaks, &network); err != nil {
return nil, err
}
if err := json.Unmarshal(resources, &h.Resources); err != nil {
@@ -80,6 +103,11 @@ func (i *Inventory) healths(ctx context.Context, only string) (map[string]NodeHe
if err := json.Unmarshal(streaks, &h.Streaks); err != nil {
return nil, fmt.Errorf("%s's health cannot be read: %w", h.Node, err)
}
if len(network) > 0 {
if err := json.Unmarshal(network, &h.Network); err != nil {
return nil, fmt.Errorf("%s's network health cannot be read: %w", h.Node, err)
}
}
out[h.Node] = h
}
return out, rows.Err()
@@ -102,18 +130,25 @@ func (i *Inventory) RecordHealth(ctx context.Context, h NodeHealth) (bool, error
if err != nil {
return false, err
}
var network []byte
if h.Network != nil {
if network, err = json.Marshal(h.Network); err != nil {
return false, err
}
}
heard := h.HeardAt
if heard.IsZero() {
heard = time.Now()
}
var node string
err = i.store.Pool().QueryRow(ctx,
`insert into node_health (node, contract, said_at, heard_at, resources, streaks)
select id, $2, $3, $4, $5, $6 from node where name = $1
`insert into node_health (node, contract, said_at, heard_at, resources, streaks, network)
select id, $2, $3, $4, $5, $6, $7 from node where name = $1
on conflict (node) do update set contract = excluded.contract, said_at = excluded.said_at,
heard_at = excluded.heard_at, resources = excluded.resources, streaks = excluded.streaks
heard_at = excluded.heard_at, resources = excluded.resources, streaks = excluded.streaks,
network = excluded.network
where node_health.said_at <= excluded.said_at
returning node`, h.Node, h.Contract, h.SaidAt, heard, resources, streaks).Scan(&node)
returning node`, h.Node, h.Contract, h.SaidAt, heard, resources, streaks, network).Scan(&node)
if errors.Is(err, pgx.ErrNoRows) {
if _, nerr := i.NodeByName(ctx, h.Node); nerr != nil {
return false, nerr
internal/inventory/migrations/0077-a-machine-says-how-its-network-is.sql
+10
View File
@@ -0,0 +1,10 @@
-- A machine says how its network is (novox/hq ADR 0241, which extends ADR 0240 from what a module runs to
-- the machine it runs on).
--
-- Beside the state of every long-running resource, each machine's node-engine states its own networking:
-- the resolver file the uplink holder declared and who rewrote it, the names through every resolver it
-- lists, the tunnel's handshake with the hub, the bus and the default route — the worst of them, since
-- when, and each part. Kept with the machine's newest statement, replaced with it, so `node show`, the
-- gate and a controller started again read the same word. Null for a machine whose node-engine is older
-- than this judging: its network is not known — never healthy, never a reason to raise anything.
alter table node_health add column network jsonb;
internal/link/protocol.go
+39
View File
@@ -428,6 +428,45 @@ type Health struct {
// At is when the engine looked, on the machine's clock: the order of its statements.
At time.Time `json:"at"`
Resources []ResourceHealth `json:"resources"`
// Network is the machine's own networking, judged by its engine (novox/hq ADR 0241); nil from an engine
// older than that judging, which is "not known", never healthy.
Network *NetworkHealth `json:"network,omitempty"`
}
// NetworkHealth is a machine's networking in one statement (ADR 0241): the worst of its parts, since
// when, and each part. The node-engine's own (mesh-host internal/link NetworkHealth).
type NetworkHealth struct {
State string `json:"state"`
Since time.Time `json:"since"`
Parts []NetworkPart `json:"parts"`
}
// The parts of a machine's networking its engine judges (ADR 0241).
const (
PartResolvConf = "resolv-conf"
PartNames = "names"
PartTunnel = "tunnel"
PartBus = "bus"
PartRoute = "route"
// TowardHub is what a part failing toward the hub names in Toward.
TowardHub = "hub"
)
// NetworkPart is one part, as the engine judged it on its second look.
type NetworkPart struct {
Part string `json:"part"`
State string `json:"state"`
// Reason is in words with no address, path or domain; Said is the detail, kept as evidence.
Reason string `json:"reason,omitempty"`
Said string `json:"said,omitempty"`
// Writer is the program that rewrote the resolver file, when the engine could name it; Owner the
// module whose file it is.
Writer string `json:"writer,omitempty"`
Owner string `json:"owner,omitempty"`
// Toward is what a failure points at: "hub", or each resolver's address that failed.
Toward []string `json:"toward,omitempty"`
Since time.Time `json:"since"`
Streak int `json:"streak,omitempty"`
}
// The states a resource is said in (ADR 0240 §4).
internal/link/protocol_test.go
+7
View File
@@ -33,6 +33,13 @@ func TestTheWireFormatIsExactlyTheseFieldNames(t *testing.T) {
[]string{"module", "resource", "kind", "target", "state", "reason", "since", "streak", "restarts",
"check", "needs"}},
{HealthSaid{Node: "n"}, []string{"node", "health"}},
// novox/hq ADR 0241: the machine's own networking, beside its resources.
{Health{Contract: ReadinessContract, Resources: []ResourceHealth{}, Network: &NetworkHealth{State: "unhealthy",
Parts: []NetworkPart{}}}, []string{"contract", "at", "resources", "network"}},
{NetworkHealth{State: "unhealthy", Parts: []NetworkPart{}}, []string{"state", "since", "parts"}},
{NetworkPart{Part: "resolv-conf", State: "unhealthy", Reason: "r", Said: "s", Writer: "w", Owner: "o",
Toward: []string{"hub"}, Streak: 2}, []string{"part", "state", "reason", "said", "writer", "owner", "toward",
"since", "streak"}},
} {
raw, err := json.Marshal(c.value)
if err != nil {