neutrino: improve perf in high latency networks

neutrino.go

@@ -753,7 +753,7 @@ func NewChainService(cfg Config) (*ChainService, error) {
 	if s.persistToDisk {
 		cfg := &chanutils.BatchWriterConfig[*filterdb.FilterData]{
 			QueueBufferSize:        chanutils.DefaultQueueSize,
-			MaxBatch:               1000,
+			MaxBatch:               100,
 			DBWritesTickerDuration: time.Millisecond * 500,
 			PutItems:               s.FilterDB.PutFilters,
 		}

query.go

@@ -49,7 +49,7 @@ var (
 	// each peer before we've concluded we aren't going to get a valid
 	// response. This allows to make up for missed messages in some
 	// instances.
-	QueryNumRetries = 2
+	QueryNumRetries = 8
 
 	// QueryPeerConnectTimeout specifies how long to wait for the
 	// underlying chain service to connect to a peer before giving up
@@ -294,7 +294,7 @@ func (s *ChainService) queryAllPeers(
 	// Starting with the set of default options, we'll apply any specified
 	// functional options to the query.
 	qo := defaultQueryOptions()
-	qo.numRetries = 1
+	qo.numRetries = 8
 	qo.applyQueryOptions(options...)
 
 	// This is done in a single-threaded query because the peerState is

rescan.go

@@ -956,7 +956,7 @@ func (rs *rescanState) handleBlockConnected(ntfn *blockntfns.Connected) error {
 
 	// Otherwise, we'll attempt to fetch the filter to retrieve the relevant
 	// transactions and notify them.
-	queryOptions := NumRetries(0)
+	queryOptions := NumRetries(2)
 	blockFilter, err := chain.GetCFilter(
 		newStamp.Hash, wire.GCSFilterRegular, queryOptions,
 	)