✨29779 source postgres slow ctid read seen on customer connection (#30125)

Co-authored-by: subodh <subodh1810@gmail.com> Co-authored-by: subodh1810 <subodh1810@users.noreply.github.com> Co-authored-by: rodireich <rodireich@users.noreply.github.com>

Author: rodireich

airbyte-integrations/connectors/source-alloydb/Dockerfile

@@ -24,5 +24,5 @@ ENV APPLICATION source-alloydb
 
 COPY --from=build /airbyte /airbyte
 
-LABEL io.airbyte.version=3.1.5
+LABEL io.airbyte.version=3.1.8
 LABEL io.airbyte.name=airbyte/source-alloydb

airbyte-integrations/connectors/source-alloydb/metadata.yaml

@@ -6,7 +6,7 @@ data:
  connectorSubtype: database
  connectorType: source
  definitionId: 1fa90628-2b9e-11ed-a261-0242ac120002
- dockerImageTag: 3.1.5
+ dockerImageTag: 3.1.8
  dockerRepository: airbyte/source-alloydb
  githubIssueLabel: source-alloydb
  icon: alloydb.svg

airbyte-integrations/connectors/source-postgres-strict-encrypt/Dockerfile

@@ -24,5 +24,5 @@ ENV APPLICATION source-postgres-strict-encrypt
 
 COPY --from=build /airbyte /airbyte
 
-LABEL io.airbyte.version=3.1.7
+LABEL io.airbyte.version=3.1.8
 LABEL io.airbyte.name=airbyte/source-postgres-strict-encrypt

airbyte-integrations/connectors/source-postgres-strict-encrypt/metadata.yaml

@@ -12,7 +12,7 @@ data:
  connectorType: source
  definitionId: decd338e-5647-4c0b-adf4-da0e75f5a750
  maxSecondsBetweenMessages: 7200
- dockerImageTag: 3.1.7
+ dockerImageTag: 3.1.8
  dockerRepository: airbyte/source-postgres-strict-encrypt
  githubIssueLabel: source-postgres
  icon: postgresql.svg

airbyte-integrations/connectors/source-postgres/Dockerfile

@@ -24,5 +24,5 @@ ENV APPLICATION source-postgres
 
 COPY --from=build /airbyte /airbyte
 
-LABEL io.airbyte.version=3.1.7
+LABEL io.airbyte.version=3.1.8
 LABEL io.airbyte.name=airbyte/source-postgres

airbyte-integrations/connectors/source-postgres/metadata.yaml

@@ -6,7 +6,7 @@ data:
  connectorSubtype: database
  connectorType: source
  definitionId: decd338e-5647-4c0b-adf4-da0e75f5a750
- dockerImageTag: 3.1.7
+ dockerImageTag: 3.1.8
  maxSecondsBetweenMessages: 7200
  dockerRepository: airbyte/source-postgres
  githubIssueLabel: source-postgres

airbyte-integrations/connectors/source-postgres/src/main/java/io/airbyte/integrations/source/postgres/PostgresQueryUtils.java

@@ -106,6 +106,15 @@ SELECT pg_relation_filenode('%s')
  WITH block_sz AS (SELECT current_setting('block_size')::int), rel_sz AS (select pg_relation_size('%s')) SELECT * from block_sz, rel_sz
  """;
 
+ /**
+ * Query estimates the max tuple in a page. We are estimating in two ways and selecting the greatest
+ * value.
+ */
+ public static final String CTID_ESTIMATE_MAX_TUPLE =
+ """
+ SELECT COALESCE(MAX((ctid::text::point)[1]::int), 0) AS max_tuple FROM "%s"."%s"
+ """;
+
  /**
  * Logs the current xmin status : 1. The number of wraparounds the source DB has undergone. (These
  * are the epoch bits in the xmin snapshot). 2. The 32-bit xmin value associated with the xmin
@@ -246,7 +255,7 @@ public static ResultWithFailed<List<io.airbyte.protocol.models.v0.AirbyteStreamN
  final List<JsonNode> jsonNodes = database.bufferedResultSetQuery(
  conn -> conn.prepareStatement(CTID_FULL_VACUUM_IN_PROGRESS_QUERY.formatted(fullTableName)).executeQuery(),
  resultSet -> JdbcUtils.getDefaultSourceOperations().rowToJson(resultSet));
- if (jsonNodes.size() != 0) {
+ if (!jsonNodes.isEmpty()) {
  Preconditions.checkState(jsonNodes.size() == 1);
  LOGGER.warn("Full Vacuum currently in progress for table {} in {} phase, the table will be skipped from syncing data", fullTableName,
  jsonNodes.get(0).get("phase"));
@@ -311,4 +320,38 @@ public static List<ConfiguredAirbyteStream> filterStreamsUnderVacuumForCtidSync(
  .toList();
  }
 
+ public static Map<AirbyteStreamNameNamespacePair, Integer> getTableMaxTupleForStreams(final JdbcDatabase database,
+ final List<ConfiguredAirbyteStream> streams,
+ final String quoteString) {
+ final Map<AirbyteStreamNameNamespacePair, Integer> tableMaxTupleEstimates = new HashMap<>();
+ streams.forEach(stream -> {
+ final AirbyteStreamNameNamespacePair namespacePair =
+ new AirbyteStreamNameNamespacePair(stream.getStream().getName(), stream.getStream().getNamespace());
+ final int maxTuple = getTableMaxTupleForStream(database, namespacePair, quoteString);
+ tableMaxTupleEstimates.put(namespacePair, maxTuple);
+ });
+ return tableMaxTupleEstimates;
+ }
+
+ public static int getTableMaxTupleForStream(final JdbcDatabase database,
+ final AirbyteStreamNameNamespacePair stream,
+ final String quoteString) {
+ try {
+ final String streamName = stream.getName();
+ final String schemaName = stream.getNamespace();
+ final String fullTableName =
+ getFullyQualifiedTableNameWithQuoting(schemaName, streamName, quoteString);
+ LOGGER.debug("running {}", CTID_ESTIMATE_MAX_TUPLE.formatted(schemaName, streamName));
+ final List<JsonNode> jsonNodes = database.bufferedResultSetQuery(
+ conn -> conn.prepareStatement(CTID_ESTIMATE_MAX_TUPLE.formatted(schemaName, streamName)).executeQuery(),
+ resultSet -> JdbcUtils.getDefaultSourceOperations().rowToJson(resultSet));
+ Preconditions.checkState(jsonNodes.size() == 1);
+ final int maxTuple = jsonNodes.get(0).get("max_tuple").asInt();
+ LOGGER.info("Stream {} max tuple is {}", fullTableName, maxTuple);
+ return maxTuple;
+ } catch (final SQLException e) {
+ throw new RuntimeException(e);
+ }
+ }
+
 }

airbyte-integrations/connectors/source-postgres/src/main/java/io/airbyte/integrations/source/postgres/PostgresSource.java

@@ -70,6 +70,7 @@
 import io.airbyte.integrations.source.postgres.ctid.CtidPerStreamStateManager;
 import io.airbyte.integrations.source.postgres.ctid.CtidPostgresSourceOperations;
 import io.airbyte.integrations.source.postgres.ctid.CtidStateManager;
+import io.airbyte.integrations.source.postgres.ctid.CtidUtils;
 import io.airbyte.integrations.source.postgres.ctid.CtidUtils.StreamsCategorised;
 import io.airbyte.integrations.source.postgres.ctid.FileNodeHandler;
 import io.airbyte.integrations.source.postgres.ctid.PostgresCtidHandler;
@@ -492,6 +493,9 @@ public List<AutoCloseableIterator<AirbyteMessage>> getIncrementalIterators(final
  finalListOfStreamsToBeSyncedViaCtid,
  getQuoteString());
 
+ final Map<io.airbyte.protocol.models.AirbyteStreamNameNamespacePair, Integer> tablesMaxTuple =
+ CtidUtils.isTidRangeScanCapableDBServer(database) ? null :
+ PostgresQueryUtils.getTableMaxTupleForStreams(database, finalListOfStreamsToBeSyncedViaCtid, getQuoteString());
  if (!streamsCategorised.ctidStreams().streamsForCtidSync().isEmpty()) {
  LOGGER.info("Streams to be synced via ctid : {}", finalListOfStreamsToBeSyncedViaCtid.size());
  LOGGER.info("Streams: {}", prettyPrintConfiguredAirbyteStreamList(finalListOfStreamsToBeSyncedViaCtid));
@@ -511,7 +515,7 @@ public List<AutoCloseableIterator<AirbyteMessage>> getIncrementalIterators(final
 
  final PostgresCtidHandler ctidHandler =
  new PostgresCtidHandler(sourceConfig, database, new CtidPostgresSourceOperations(Optional.empty()), getQuoteString(),
- fileNodeHandler, tableBlockSizes, ctidStateManager,
+ fileNodeHandler, tableBlockSizes, tablesMaxTuple, ctidStateManager,
  namespacePair -> Jsons.jsonNode(xminStatus));
 
  final List<AutoCloseableIterator<AirbyteMessage>> initialSyncCtidIterators = new ArrayList<>(ctidHandler.getInitialSyncCtidIterator(
@@ -557,6 +561,11 @@ public List<AutoCloseableIterator<AirbyteMessage>> getIncrementalIterators(final
  database,
  finalListOfStreamsToBeSyncedViaCtid,
  getQuoteString());
+
+ final Map<io.airbyte.protocol.models.AirbyteStreamNameNamespacePair, Integer> tablesMaxTuple =
+ CtidUtils.isTidRangeScanCapableDBServer(database) ? null :
+ PostgresQueryUtils.getTableMaxTupleForStreams(database, finalListOfStreamsToBeSyncedViaCtid, getQuoteString());
+
  if (finalListOfStreamsToBeSyncedViaCtid.isEmpty()) {
  LOGGER.info("No Streams will be synced via ctid.");
  } else {
@@ -581,6 +590,7 @@ public List<AutoCloseableIterator<AirbyteMessage>> getIncrementalIterators(final
  getQuoteString(),
  fileNodeHandler,
  tableBlockSizes,
+ tablesMaxTuple,
  ctidStateManager,
  namespacePair -> Jsons.jsonNode(cursorBasedStatusMap.get(namespacePair)));
 

airbyte-integrations/connectors/source-postgres/src/main/java/io/airbyte/integrations/source/postgres/cdc/PostgresCdcCtidInitializer.java

@@ -26,6 +26,7 @@
 import io.airbyte.integrations.source.postgres.ctid.CtidPostgresSourceOperations;
 import io.airbyte.integrations.source.postgres.ctid.CtidPostgresSourceOperations.CdcMetadataInjector;
 import io.airbyte.integrations.source.postgres.ctid.CtidStateManager;
+import io.airbyte.integrations.source.postgres.ctid.CtidUtils;
 import io.airbyte.integrations.source.postgres.ctid.FileNodeHandler;
 import io.airbyte.integrations.source.postgres.ctid.PostgresCtidHandler;
 import io.airbyte.integrations.source.relationaldb.TableInfo;
@@ -140,11 +141,17 @@ public static List<AutoCloseableIterator<AirbyteMessage>> cdcCtidIteratorsCombin
  database,
  finalListOfStreamsToBeSyncedViaCtid,
  quoteString);
+
+ final Map<io.airbyte.protocol.models.AirbyteStreamNameNamespacePair, Integer> tablesMaxTuple =
+ CtidUtils.isTidRangeScanCapableDBServer(database) ? null :
+ PostgresQueryUtils.getTableMaxTupleForStreams(database, finalListOfStreamsToBeSyncedViaCtid, quoteString);
+
  final PostgresCtidHandler ctidHandler = new PostgresCtidHandler(sourceConfig, database,
  ctidPostgresSourceOperations,
  quoteString,
  fileNodeHandler,
  tableBlockSizes,
+ tablesMaxTuple,
  ctidStateManager,
  namespacePair -> Jsons.emptyObject());
 

airbyte-integrations/connectors/source-postgres/src/main/java/io/airbyte/integrations/source/postgres/ctid/Ctid.java

@@ -16,6 +16,7 @@ public class Ctid {
 
  final Long page;
  final Long tuple;
+ public static final Ctid ZERO = Ctid.of(0, 0);
 
  public static Ctid of(final long page, final long tuple) {
  return new Ctid(page, tuple);
@@ -71,4 +72,7 @@ public int hashCode() {
  return Objects.hash(page, tuple);
  }
 
+ public static Ctid inc(final Ctid ctid, final long maxTuple) {
+ return (ctid.tuple + 1 > maxTuple) ? Ctid.of(ctid.page + 1, 1) : Ctid.of(ctid.page, ctid.tuple + 1);
+ }
 }