cache/disk: census counts an eviction as never-read from its access state, so boot-scanned entries stop inflating evicted_never_read after a restart

cache/disk/census.go

@@ -425,7 +425,6 @@ func (r *censusRecorder) OnEvict(key string, value lruItem) {
 	if now > value.addedAt {
 		createdAge = now - value.addedAt
 	}
-	neverRead := value.lastAccess == value.addedAt
 	lastReadIdx := censusBucketIdx(lastReadAge)
 	createdIdx := censusBucketIdx(createdAge)
 
@@ -436,13 +435,16 @@ func (r *censusRecorder) OnEvict(key string, value lruItem) {
 	acc.residentEntries--
 	acc.evictedBytes += value.sizeOnDisk
 	acc.evictedEntries++
-	if neverRead {
+	switch value.accessState {
+	case itemStateNew:
 		acc.evictedNeverReadByes += value.sizeOnDisk
-	} else {
+	case itemStateTouched:
 		if lastReadAge < censusLiveThreshold {
 			acc.evictedLiveBytes += value.sizeOnDisk
 		}
 		acc.evictedByLastReadAge[lastReadIdx] += value.sizeOnDisk
+		// itemStateBootScanned: lastAccess is file atime (frozen at creation on
+		// noatime mounts), not a read, so the entry has no last-read history.
 	}
 	acc.evictedByCreatedAge[createdIdx] += value.sizeOnDisk
 	acc.removeResident(class, value.sizeOnDisk, value.accessState)
@@ -452,16 +454,15 @@ func (r *censusRecorder) OnEvict(key string, value lruItem) {
 	r.mu.Unlock()
 
 	sz := float64(value.sizeOnDisk)
-	// Never-read entries have no last read: they are counted in the
-	// never-read counter and the created-age buckets only, so the
-	// last_read series stays a pure harm signal (bytes someone actually
-	// used, evicted anyway). This also gives thrash alerts built-in
-	// restart tolerance: boot-scanned entries look never-read until first
-	// touched, so a roll cannot fabricate still-live evictions.
-	if neverRead {
+	// Only touched entries have a last read, so the last_read series stays
+	// a pure harm signal (bytes someone actually used, evicted anyway) and
+	// a roll cannot fabricate still-live evictions out of boot-scanned
+	// entries; those count in the created-age buckets only.
+	switch value.accessState {
+	case itemStateNew:
 		r.evictedNeverReadBytes.WithLabelValues(tool).Add(sz)
 		r.evictedNeverReadEntries.WithLabelValues(tool).Inc()
-	} else {
+	case itemStateTouched:
 		r.evictedBytesByAge.WithLabelValues("last_read", censusBucketLabels[lastReadIdx], tool).Add(sz)
 		r.evictedEntriesByAge.WithLabelValues("last_read", censusBucketLabels[lastReadIdx], tool).Inc()
 	}

cache/disk/census_test.go

@@ -409,6 +409,28 @@ func TestCensusBootScannedIsUnknownNotNeverRead(t *testing.T) {
 	if rows[0].EvictedNeverReadBytesByClass[class] != 0 {
 		t.Errorf("EvictedNeverReadBytesByClass = %d, want 0", rows[0].EvictedNeverReadBytesByClass[class])
 	}
+	// The scalar and its counters follow the same access state: a boot-scanned
+	// entry has addedAt == lastAccess, which must not read as never-read.
+	if rows[0].EvictedNeverReadBytes != 0 || rows[0].EvictedBytes != 50 {
+		t.Errorf("EvictedNeverReadBytes/EvictedBytes = %d/%d, want 0/50", rows[0].EvictedNeverReadBytes, rows[0].EvictedBytes)
+	}
+	if got := testutil.ToFloat64(rec.evictedNeverReadEntries.WithLabelValues("unknown")); got != 0 {
+		t.Errorf("evictedNeverReadEntries = %v, want 0", got)
+	}
+	// Nor as read: its lastAccess is file atime, not a hit.
+	if rows[0].EvictedLiveBytes != 0 {
+		t.Errorf("EvictedLiveBytes = %d, want 0", rows[0].EvictedLiveBytes)
+	}
+	for i, v := range rows[0].EvictedByLastReadAge {
+		if v != 0 {
+			t.Errorf("EvictedByLastReadAge[%d] = %d, want 0", i, v)
+		}
+	}
+	for _, label := range censusBucketLabels {
+		if got := testutil.ToFloat64(rec.evictedEntriesByAge.WithLabelValues("last_read", label, "unknown")); got != 0 {
+			t.Errorf("evictedEntriesByAge last_read %s = %v, want 0", label, got)
+		}
+	}
 	if rows[0].ResidentUnknownBytesByClass[class] != 0 || rows[0].ResidentBytesByClass[class] != 50 {
 		t.Errorf("resident unknown/total = %d/%d, want 0/50",
 			rows[0].ResidentUnknownBytesByClass[class], rows[0].ResidentBytesByClass[class])