Files
felhom.eu/hub/internal/monitor/deadline_tiers_test.go
T
Claude Code b11607b26b hub v0.76.0 — R-82 Slice C: tier-aware backup thresholds
R-81 merged every backup signal into one 'newest' against a single 26h limit.
backupStaleAfter's own comment recorded why that stops being right under a
weekly offsite tier. Each tier is now judged against its own threshold;
R-81's structure (three verdicts, anchored absence, distinct reasons) and its
boundary test are preserved intact.

- offsiteBackupStaleAfter = 8d (7d cadence + headroom); backupStaleAfter keeps
  26h and now names the HOST tier only
- splitTiers / assessTier / newestBackupEvidenceByTier

Slice-A.4 rule implemented: a PBS-targeted vzdump appears in BOTH arrays, so
classification is by TARGET TYPE (target_id -> storage_targets[].name -> type),
never by array membership — otherwise a PBS backup makes a stale host tier look
fresh. storage_targets is used rather than pbs_dr.storage_id because the latter
is null on a box with a PBS storage but no DR descriptor.

A tier is only judged when the box HAS it, else every box without an offsite
tier would alarm once the anchor elapsed — R-81's mistake one level down. With
neither tier identifiable (old agent) the pre-Slice-C path runs unchanged.

Intended behaviour change: a 30h offsite snapshot no longer alarms. Three
fixtures asserted the merged threshold; each still asserts an alarm at the
correct limit. No assertion was weakened.

RECORDED LIMITATION: the hub infers 'PBS => weekly' from storage type.
defaultBackupTarget is felhom-pbs, so a box that never sets local_backup_target
would run PBS as its DAILY tier and be judged against 8 days — 7 days of
blindness. No box is in that shape today; the real fix is the agent reporting
per-tier cadences. Own task.

Red-proof observed. Replayed live: demo-felhom OK, demo-hp UNKNOWN (defers
correctly), drill-r50 MISSED (true positive). No customer email would be sent.
2026-07-26 16:58:38 +02:00

193 lines
8.1 KiB
Go

package monitor
import (
"encoding/json"
"strings"
"testing"
"time"
)
// R-82 Slice C — tier-aware thresholds.
//
// The whole point: a healthy WEEKLY offsite snapshot is >26h old six days in seven, so the merged
// threshold would cry wolf on it — exactly what `backupStaleAfter`'s comment predicted. Each tier is
// now judged against its own limit, and R-81's structure (three verdicts, anchored absence, one
// distinct reason per failure mode) is preserved intact.
func tieredReport(pbsAge time.Duration, verify string, hostAge time.Duration, now time.Time) string {
at := func(d time.Duration) string { return now.Add(-d).Format(time.RFC3339) }
return `{
"storage_targets":[
{"name":"local","type":"local","content":"backup,iso"},
{"name":"felhom-pbs","type":"pbs","content":"backup"}
],
"pbs_snapshots":[{"backup_time":"` + at(pbsAge) + `","verify_state":"` + verify + `"}],
"backups":[{"target_id":"local","started_at":"` + at(hostAge) + `","success":true}]
}`
}
func freshEv(now time.Time, hostAge, offAge time.Duration) backupEvidence {
return backupEvidence{
newestHost: now.Add(-hostAge), haveHost: true,
newestOffsite: now.Add(-offAge), haveOffsite: true,
firstReportAt: now.Add(-60 * 24 * time.Hour),
}
}
// ── THE SCENARIO ─────────────────────────────────────────────────────────────────────────────
//
// A 6-day-old offsite snapshot with a fresh host backup is SILENT; the same snapshot at 9 days
// ALARMS.
//
// COMPANION RED-PROOF (observed): give tierOffsite the host threshold
// (`tierOffsite = backupTier{name:"offsite", staleWhen: backupStaleAfter}`) — i.e. restore the
// single merged threshold — and the 6-day case fails with
//
// 6-day-old offsite snapshot is HEALTHY under a weekly cadence and must be SILENT;
// got verdict=2 reason="offsite tier: newest backup is 144h0m0s old (limit 26h0m0s)"
//
// which is verbatim the cry-wolf this slice exists to remove. Restored.
func TestTierAware_SixDayOffsiteSilent_NineDayAlarms(t *testing.T) {
now := time.Date(2026, 7, 26, 3, 0, 0, 0, time.UTC)
t.Run("6 days offsite + fresh host → SILENT", func(t *testing.T) {
got := assessBackupFreshness(tieredReport(6*24*time.Hour, "ok", 2*time.Hour, now),
freshEv(now, 2*time.Hour, 6*24*time.Hour), now)
if got.missed() {
t.Fatalf("6-day-old offsite snapshot is HEALTHY under a weekly cadence and must be SILENT; got verdict=%d reason=%q",
got.verdict, got.reason)
}
if got.verdict != verdictOK {
t.Fatalf("both tiers have evidence inside their limits → OK, not a deferred UNKNOWN; got verdict=%d reason=%q", got.verdict, got.reason)
}
})
t.Run("9 days offsite + fresh host → ALARMS, naming the offsite tier", func(t *testing.T) {
got := assessBackupFreshness(tieredReport(9*24*time.Hour, "ok", 2*time.Hour, now),
freshEv(now, 2*time.Hour, 9*24*time.Hour), now)
if !got.missed() {
t.Fatalf("9-day-old offsite snapshot is past the 8-day limit and MUST alarm; got verdict=%d reason=%q",
got.verdict, got.reason)
}
if !strings.HasPrefix(got.reason, "offsite tier:") {
t.Fatalf("the reason must name the OFFSITE tier — a merged message would have made the R-80 diagnosis impossible; got %q", got.reason)
}
if strings.Contains(got.reason, "host tier") {
t.Fatalf("the host tier is fresh and must not appear in the reason; got %q", got.reason)
}
})
}
// Neither tier rescues the other. This is the merged-threshold bug in both directions.
func TestTierAware_TiersDoNotRescueEachOther(t *testing.T) {
now := time.Date(2026, 7, 26, 3, 0, 0, 0, time.UTC)
t.Run("fresh host does not rescue a stale offsite tier", func(t *testing.T) {
got := assessBackupFreshness(tieredReport(9*24*time.Hour, "ok", time.Hour, now),
freshEv(now, time.Hour, 9*24*time.Hour), now)
if !got.missed() {
t.Fatalf("a fresh DAILY backup must not satisfy the WEEKLY tier; got %q", got.reason)
}
})
t.Run("fresh offsite does not rescue a stale host tier", func(t *testing.T) {
got := assessBackupFreshness(tieredReport(time.Hour, "ok", 30*time.Hour, now),
freshEv(now, 30*time.Hour, time.Hour), now)
if !got.missed() {
t.Fatalf("a fresh WEEKLY snapshot must not satisfy the DAILY tier; got %q", got.reason)
}
if !strings.Contains(got.reason, "host tier") {
t.Fatalf("the reason must name the HOST tier; got %q", got.reason)
}
})
t.Run("both stale → BOTH named", func(t *testing.T) {
got := assessBackupFreshness(tieredReport(9*24*time.Hour, "ok", 30*time.Hour, now),
freshEv(now, 30*time.Hour, 9*24*time.Hour), now)
if !got.missed() {
t.Fatalf("both tiers stale must alarm; got %q", got.reason)
}
if !strings.Contains(got.reason, "host tier") || !strings.Contains(got.reason, "offsite tier") {
t.Fatalf("a box with TWO broken tiers must not report only one; got %q", got.reason)
}
})
}
// ── The A.4 rule: classify by TARGET TYPE, never by array membership ─────────────────────────
//
// A PBS-targeted vzdump appears in BOTH arrays. Treating every `backups[]` entry as the host tier
// would let a PBS backup make a STALE HOST TIER LOOK FRESH — silently losing the daily tier's alarm.
func TestTierAware_PBSTargetedVzdumpIsNotHostEvidence(t *testing.T) {
now := time.Date(2026, 7, 26, 3, 0, 0, 0, time.UTC)
at := func(d time.Duration) string { return now.Add(-d).Format(time.RFC3339) }
// The host tier's own backup is 40h old (stale). A FRESH PBS-targeted vzdump sits in the same
// array — exactly the double-appearance Slice A.4 recorded.
report := `{
"storage_targets":[
{"name":"local","type":"local","content":"backup"},
{"name":"felhom-pbs","type":"pbs","content":"backup"}
],
"pbs_snapshots":[{"backup_time":"` + at(time.Hour) + `","verify_state":"ok"}],
"backups":[
{"target_id":"local","started_at":"` + at(40*time.Hour) + `","success":true},
{"target_id":"felhom-pbs","started_at":"` + at(time.Hour) + `","success":true}
]
}`
ev := backupEvidence{
newestHost: now.Add(-40 * time.Hour), haveHost: true,
newestOffsite: now.Add(-time.Hour), haveOffsite: true,
firstReportAt: now.Add(-60 * 24 * time.Hour),
}
got := assessBackupFreshness(report, ev, now)
if !got.missed() {
t.Fatalf("a PBS-targeted vzdump must NOT count as host-tier evidence — the host tier is 40h stale; got verdict=%d reason=%q",
got.verdict, got.reason)
}
if !strings.Contains(got.reason, "host tier") {
t.Fatalf("the stale HOST tier must be named; got %q", got.reason)
}
}
// splitTiers is the classifier the whole slice rests on.
func TestSplitTiers_Classification(t *testing.T) {
now := time.Date(2026, 7, 26, 3, 0, 0, 0, time.UTC)
at := func(d time.Duration) string { return now.Add(-d).Format(time.RFC3339) }
t.Run("no storage_targets and no evidence → neither tier expected (old-agent fallback)", func(t *testing.T) {
var hr hostReportBackups
mustUnmarshal(t, `{"pbs_snapshots":[],"backups":[]}`, &hr)
h, o := splitTiers(hr)
if h.expected || o.expected {
t.Fatalf("nothing identifiable must leave BOTH tiers unexpected so the pre-Slice-C path runs; got host=%v offsite=%v", h.expected, o.expected)
}
})
t.Run("a PBS storage alone makes the offsite tier expected", func(t *testing.T) {
var hr hostReportBackups
mustUnmarshal(t, `{"storage_targets":[{"name":"felhom-pbs","type":"pbs","content":"backup"}]}`, &hr)
_, o := splitTiers(hr)
if !o.expected || o.have {
t.Fatalf("a configured-but-empty offsite tier is EXPECTED with no evidence — that is the R-82 fault; got %+v", o)
}
})
t.Run("verify_state failed is carried on the newest offsite snapshot only", func(t *testing.T) {
var hr hostReportBackups
mustUnmarshal(t, `{"pbs_snapshots":[
{"backup_time":"`+at(48*time.Hour)+`","verify_state":"failed"},
{"backup_time":"`+at(time.Hour)+`","verify_state":"ok"}]}`, &hr)
_, o := splitTiers(hr)
if o.failed {
t.Fatalf("an OLDER failed snapshot must not condemn a newer good one; got %+v", o)
}
})
}
func mustUnmarshal(t *testing.T, s string, v any) {
t.Helper()
if err := json.Unmarshal([]byte(s), v); err != nil {
t.Fatal(err)
}
}