historyarchive: Improve existence checks and performance (#5179)

historyarchive/archive_cache.go

@@ -113,7 +113,26 @@ func (abc *ArchiveBucketCache) GetFile(
 }
 
 func (abc *ArchiveBucketCache) Exists(filepath string) bool {
-	return abc.lru.Contains(path.Join(abc.path, filepath))
+	localPath := path.Join(abc.path, filepath)
+
+	// First, check if the file exists in the cache.
+	if abc.lru.Contains(localPath) {
+		return true
+	}
+
+	// If it doesn't, it may still exist on the disk which is still a cheaper
+	// check than going upstream.
+	//
+	// Note that this means the cache and disk are out of sync (perhaps due to
+	// other archives using the same cache location) so we can update it. This
+	// situation is well-handled by `GetFile`.
+	_, statErr := os.Stat(localPath)
+	if statErr == nil || os.IsExist(statErr) {
+		abc.lru.Add(localPath, struct{}{})
+		return true
+	}
+
+	return false
 }
 
 // Close purges the cache and cleans up the filesystem.

ingest/verify/main.go

@@ -66,7 +66,7 @@ func (v *StateVerifier) GetLedgerEntries(count int) ([]xdr.LedgerEntry, error) {
 	}
 
 	entries := make([]xdr.LedgerEntry, 0, count)
-	v.currentEntries = make(map[string]xdr.LedgerEntry)
+	v.currentEntries = make(map[string]xdr.LedgerEntry, count)
 
 	for count > 0 {
 		entryChange, err := v.stateReader.Read()

services/horizon/internal/ingest/verify.go

@@ -157,8 +157,8 @@ func (s *system) verifyState(verifyAgainstLatestCheckpoint bool) error {
 				}
 			}
 		}
-		log.WithField("duration", duration).Info("State verification finished")
 
+		localLog.WithField("duration", duration).Info("State verification finished")
 	}()
 
 	localLog.Info("Creating state reader...")