[
https://issues.apache.org/jira/browse/PHOENIX-6420?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17308867#comment-17308867
]
ASF GitHub Bot commented on PHOENIX-6420:
-----------------------------------------
tkhurana commented on a change in pull request #1183:
URL: https://github.com/apache/phoenix/pull/1183#discussion_r601719128
##########
File path:
phoenix-core/src/main/java/org/apache/phoenix/execute/MutationState.java
##########
@@ -404,67 +434,89 @@ public int getNumRows() {
return numRows;
}
+ private MultiRowMutationState getLastMutationBatch(Map<TableRef,
List<MultiRowMutationState>> mutations, TableRef tableRef) {
+ List<MultiRowMutationState> mutationBatches = mutations.get(tableRef);
+ if (mutationBatches == null || mutationBatches.isEmpty()) {
+ return null;
+ }
+ return mutationBatches.get(mutationBatches.size() - 1);
+ }
+
private void joinMutationState(TableRef tableRef, MultiRowMutationState
srcRows,
- Map<TableRef, MultiRowMutationState> dstMutations) {
+ Map<TableRef, List<MultiRowMutationState>> dstMutations) {
PTable table = tableRef.getTable();
boolean isIndex = table.getType() == PTableType.INDEX;
- boolean incrementRowCount = dstMutations == this.mutations;
- MultiRowMutationState existingRows = dstMutations.put(tableRef,
srcRows);
- if (existingRows != null) { // Rows for that table already exist
- // Loop through new rows and replace existing with new
- for (Map.Entry<ImmutableBytesPtr, RowMutationState> rowEntry :
srcRows.entrySet()) {
- // Replace existing row with new row
- RowMutationState existingRowMutationState =
existingRows.put(rowEntry.getKey(), rowEntry.getValue());
- if (existingRowMutationState != null) {
- Map<PColumn, byte[]> existingValues =
existingRowMutationState.getColumnValues();
- if (existingValues != PRow.DELETE_MARKER) {
- Map<PColumn, byte[]> newRow =
rowEntry.getValue().getColumnValues();
- // if new row is PRow.DELETE_MARKER, it means delete,
and we don't need to merge it with
- // existing row.
- if (newRow != PRow.DELETE_MARKER) {
- // decrement estimated size by the size of the old
row
- estimatedSize -=
existingRowMutationState.calculateEstimatedSize();
- // Merge existing column values with new column
values
- existingRowMutationState.join(rowEntry.getValue());
- // increment estimated size by the size of the new
row
- estimatedSize +=
existingRowMutationState.calculateEstimatedSize();
- // Now that the existing row has been merged with
the new row, replace it back
- // again (since it was merged with the new one
above).
- existingRows.put(rowEntry.getKey(),
existingRowMutationState);
- }
- }
- } else {
- if (incrementRowCount && !isIndex) { // Don't count index
rows in row count
- numRows++;
- // increment estimated size by the size of the new row
- estimatedSize +=
rowEntry.getValue().calculateEstimatedSize();
- }
- }
- }
- // Put the existing one back now that it's merged
- dstMutations.put(tableRef, existingRows);
- } else {
+ boolean incrementRowCount = dstMutations == this.mutationsMap;
+ // we only need to check if the new mutation batch (srcRows) conflicts
with the
+ // last mutation batch
Review comment:
Because we only try to merge the new batch with the last one. That is
why we only check for conflicts with the last one. All previous batches are not
updated.
--
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.
For queries about this service, please contact Infrastructure at:
[email protected]
> Wrong result when conditional and regular upserts are passed in the same
> commit batch
> -------------------------------------------------------------------------------------
>
> Key: PHOENIX-6420
> URL: https://issues.apache.org/jira/browse/PHOENIX-6420
> Project: Phoenix
> Issue Type: Bug
> Reporter: Tanuj Khurana
> Assignee: Tanuj Khurana
> Priority: Major
>
> Consider this example:
> {code:java}
> CREATE TABLE T1 (k integer not null primary key, v1 bigint, v2 bigint);
> {code}
> Now consider this batch:
> {code:java}
> UPSERT INTO T1 VALUES(0,0,1);
> UPSERT INTO T1 VALUES(0,1,1) ON DUPLICATE KEY UPDATE v1 = v1 + 2;
> commit();
> {code}
> Expected row state: 0, 2, 1
> Actual: 0, 2, 0
> The value of the column (v2) not updated in the conditional expression
> remains default. It's value should have been the one set in the regular
> upsert in the batch.
> Now, the row exists. Consider another batch of updates
> {code:java}
> UPSERT INTO T1 VALUES(0, 7, 4);
> UPSERT INTO T1 VALUES(0,1,1) ON DUPLICATE KEY UPDATE v1 = v1 + 2;
> commit();
> {code}
> Expected row state: 0,2,1 -> 0, 9, 4
> Actual: 0,2,0 -> 0, 4, 0
> The conditional update expression is evaluated and applied on the row state
> already committed instead of on the regular update in the same batch. Also,
> v2 still remains 0 (the default value).
> Now consider the case of a partial regular update following a conditional
> update:
> {code:java}
> UPSERT INTO T1 (k, v2) VALUES(0,100) ON DUPLICATE KEY UPDATE v1 = v1 + 2;
> UPSERT INTO T1 (k, v2) VALUES (0,125);
> commit();
> {code}
> Expected row state: 0, 9, 4 -> 0, 11, 125
> Actual: 0, 4, 0 -> 0, 4, 125
> Only the regular update is applied and the conditional update is completely
> ignored.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
