vldpyatkov commented on code in PR #2475:
URL: https://github.com/apache/ignite-3/pull/2475#discussion_r1302741325
##########
modules/table/src/main/java/org/apache/ignite/internal/table/distributed/replicator/PartitionReplicaListener.java:
##########
@@ -970,28 +976,41 @@ private CompletableFuture<Void> continueIndexScan(
return completedFuture(null); // End of range reached.
Exit loop.
}
- return lockManager.acquire(txId, new LockKey(tableId(),
currentRow.rowId()), LockMode.S)
+ RowId rowId = currentRow.rowId();
+
+ return lockManager.acquire(txId, new LockKey(tableId(),
rowId), LockMode.S)
.thenComposeAsync(rowLock -> { // Table row S lock
- ReadResult readResult =
mvDataStorage.read(currentRow.rowId(), HybridTimestamp.MAX_VALUE);
- return
resolveAndCheckReadCompatibility(readResult, txId)
- .thenCompose(resolvedReadResult -> {
- if (resolvedReadResult != null) {
- if
(indexRowMatches(currentRow, resolvedReadResult, schemaAwareIndexStorage)) {
-
result.add(resolvedReadResult);
- }
- }
-
- // Proceed scan.
- return continueIndexScan(
- txId,
- schemaAwareIndexStorage,
- indexLocker,
- indexCursor,
- batchSize,
- result,
- isUpperBoundAchieved
- );
- });
+ ReadResult readResult =
mvDataStorage.read(rowId, HybridTimestamp.MAX_VALUE);
+ return
resolveAndCheckReadCompatibility(readResult, txId, () -> {
+ if (readResult.newestCommitTimestamp() ==
null) {
Review Comment:
Because here we can be sure that is a write-intent. It happens after the
resolution is finalized and the decision about getting committed value is
applied.
##########
modules/table/src/main/java/org/apache/ignite/internal/table/distributed/replicator/PartitionReplicaListener.java:
##########
@@ -1025,15 +1044,26 @@ private CompletableFuture<Void> continueIndexLookup(
return lockManager.acquire(txId, new LockKey(tableId(), rowId),
LockMode.S)
.thenComposeAsync(rowLock -> { // Table row S lock
ReadResult readResult = mvDataStorage.read(rowId,
HybridTimestamp.MAX_VALUE);
- return resolveAndCheckReadCompatibility(readResult, txId)
- .thenCompose(resolvedReadResult -> {
- if (resolvedReadResult != null) {
- result.add(resolvedReadResult);
- }
+ return resolveAndCheckReadCompatibility(readResult, txId,
() -> {
+ if (readResult.newestCommitTimestamp() == null) {
Review Comment:
We have a different resolution procedure for cursors and for storage values.
If we hide this for storage, we will check it for cursors twice.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
