Status says when each machine last reported, beside when it was sent

"Not waiting" says the declaration is current, not that the machine
finished applying it: the sent digest is recorded at send. So a test
that pushed, saw waiting clear, and asked the machine what it was
running found containers that did not exist yet — the certificate fix
made compositions stable, and the settling that used to fail first had
been hiding the gap behind it.

The mesh already held the missing half: every machine's last report,
with its time. It just was not in the JSON. `reported` now sets each
machine's last word beside when the current declaration went to it, and
"has it caught up" becomes a comparison of two timestamps the mesh
recorded itself — a report newer than the send means the machine acted
on what was sent; older means it is still working, which waiting alone
cannot distinguish.
This commit is contained in:
2026-09-01 23:02:26 +02:00
parent ef7750f816
commit c0107b8572
6 changed files with 69 additions and 6 deletions
+1 -1
View File
@@ -95,7 +95,7 @@ func board() http.Handler {
return
}
body, err := statusAsJSON(asked.wrong, asked.nodes, asked.quiet, asked.behind, asked.sources,
asked.waiting)
asked.waiting, asked.reported)
if err != nil {
http.Error(w, err.Error(), http.StatusInternalServerError)
return
+4
View File
@@ -438,4 +438,8 @@ type answers struct {
sources map[string]inventory.Source
// waiting is every machine not running what the mesh would send it.
waiting []inventory.Machine
// reported is every machine's last word beside when it was last sent a declaration — the
// pair that answers "has it caught up", which waiting alone cannot (the sent digest is
// recorded at send, not at apply).
reported []inventory.Reported
}
+19 -2
View File
@@ -35,6 +35,11 @@ type meshStatus struct {
// Behind one level down: that says the catalogue is old, this says a machine is — and only
// this one has somebody's change waiting inside it.
Waiting []machineWaiting `json:"waiting"`
// Reported is every machine's last word beside when it was last sent a declaration. A
// machine whose report is newer than its send has acted on the current declaration; one
// whose is older is still working — and Waiting cannot tell those apart, because the sent
// digest is recorded at send, not at apply.
Reported []machineReported `json:"reported"`
// Machines is how many the mesh knows about, so a reader can tell "none wrong" from
// "none at all".
Machines int `json:"machines"`
@@ -54,6 +59,13 @@ type machineDoing struct {
At time.Time `json:"at"`
}
type machineReported struct {
Node string `json:"node"`
Outcome string `json:"outcome"`
At *time.Time `json:"at,omitempty"`
Sent *time.Time `json:"sent,omitempty"`
}
type machineWaiting struct {
Node string `json:"node"`
// Never is true when nothing has ever been sent to it. Not out of date: nobody has ever asked
@@ -79,10 +91,15 @@ type moduleBehind struct {
// statusAsJSON answers the same three questions as the text form, from the same calls.
func statusAsJSON(wrong []inventory.Doing, nodes []inventory.Node, quiet []inventory.Node,
behind map[string][]string, sources map[string]inventory.Source,
waiting []inventory.Machine) ([]byte, error) {
waiting []inventory.Machine, reported []inventory.Reported) ([]byte, error) {
out := meshStatus{Machines: len(nodes), Wrong: []machineDoing{},
Quiet: []machineQuiet{}, Behind: []moduleBehind{}, Waiting: []machineWaiting{}}
Quiet: []machineQuiet{}, Behind: []moduleBehind{}, Waiting: []machineWaiting{},
Reported: []machineReported{}}
for _, r := range reported {
out.Reported = append(out.Reported, machineReported{
Node: r.Node, Outcome: r.Outcome, At: r.At, Sent: r.Sent})
}
for _, m := range waiting {
out.Waiting = append(out.Waiting, machineWaiting{
+2 -2
View File
@@ -17,7 +17,7 @@ import (
func statusOf(t *testing.T, wrong []inventory.Doing, nodes, quiet []inventory.Node,
behind map[string][]string, sources map[string]inventory.Source) map[string]any {
t.Helper()
body, err := statusAsJSON(wrong, nodes, quiet, behind, sources, nil)
body, err := statusAsJSON(wrong, nodes, quiet, behind, sources, nil, nil)
if err != nil {
t.Fatal(err)
}
@@ -126,7 +126,7 @@ func TestNoSecretIsInWhatABoardReads(t *testing.T) {
body, err := statusAsJSON(
[]inventory.Doing{{Node: "a", Outcome: inventory.OutcomeRefused,
Refused: "resource \"x\": a file needs a path"}},
[]inventory.Node{{Name: "a"}}, nil, nil, nil, nil)
[]inventory.Node{{Name: "a"}}, nil, nil, nil, nil, nil)
if err != nil {
t.Fatal(err)
}
+6 -1
View File
@@ -54,7 +54,8 @@ func statusCommand(ctx context.Context, args []string) error {
behind, sources := asked.behind, asked.sources
if *asJSON {
body, err := statusAsJSON(wrong, nodes, quiet, behind, sources, asked.waiting)
body, err := statusAsJSON(wrong, nodes, quiet, behind, sources, asked.waiting,
asked.reported)
if err != nil {
return err
}
@@ -207,6 +208,10 @@ func theThreeQuestions(ctx context.Context, open *stores) (answers, error) {
if err != nil {
return answers{}, err
}
out.reported, err = inv.LastReports(ctx)
if err != nil {
return answers{}, err
}
out.sources = map[string]inventory.Source{}
for module := range out.behind {
from, err := inv.SourceOf(ctx, module)
+37
View File
@@ -514,6 +514,43 @@ func (i *Inventory) DoingOf(ctx context.Context, name string) (Doing, bool, erro
return d, true, nil
}
// Reported is one machine's last word set beside what was last asked of it.
//
// The pair is what makes "has it caught up" answerable: a machine that reported *after* it was
// sent the current declaration has acted on it; one that reported before is still working, or
// has not started — and "not waiting" alone cannot tell those apart, because the sent digest is
// recorded at send, not at apply. The lab asserted on a machine the moment its declaration was
// current and found containers that did not exist yet.
type Reported struct {
Node string
Outcome string
// At is when it last reported; nil if it never has.
At *time.Time
// Sent is when the current declaration went to it; nil if nothing ever did.
Sent *time.Time
}
// LastReports is every machine's last report beside when it was last sent a declaration.
func (i *Inventory) LastReports(ctx context.Context) ([]Reported, error) {
rows, err := i.store.Pool().Query(ctx,
`select n.name, coalesce(r.outcome, ''), r.at, n.sent_at
from node n left join node_report r on r.node = n.id
order by n.name`)
if err != nil {
return nil, err
}
defer rows.Close()
var out []Reported
for rows.Next() {
var r Reported
if err := rows.Scan(&r.Node, &r.Outcome, &r.At, &r.Sent); err != nil {
return nil, err
}
out = append(out, r)
}
return out, rows.Err()
}
// RecordSent keeps a digest of the declaration a machine was last sent.
//
// **A digest rather than the declaration.** The mesh can compute what a machine should be at any