[
https://issues.apache.org/jira/browse/PHOENIX-7004?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17780504#comment-17780504
]
ASF GitHub Bot commented on PHOENIX-7004:
-----------------------------------------
virajjasani commented on PR #1724:
URL: https://github.com/apache/phoenix/pull/1724#issuecomment-1783362587
> @virajjasani Thinking what changed from this PR #1697 ?
this is the main diff:
```
public static void verifyKeyInScanRange(ImmutableBytesWritable ptr,
Scan scan,
Tuple tuple)
throws ResultSetOutOfScanRangeException {
if ((tuple.size() == 1 && tuple.getValue(SINGLE_COLUMN_FAMILY,
SINGLE_COLUMN) != null)
|| (tuple.size() == 1 && tuple.getValue(OFFSET_FAMILY,
OFFSET_COLUMN) != null)
|| isDummy(tuple)) {
return;
}
if (isLocalIndex(scan)) {
verifyScanRanges(ptr, scan,
scan.getAttribute(SCAN_START_ROW_SUFFIX),
scan.getAttribute(SCAN_STOP_ROW_SUFFIX));
} else {
try {
verifyScanRanges(ptr, scan, scan.getStartRow(),
scan.getStopRow());
} catch (ResultSetOutOfScanRangeException e) {
// required for Salt Buckets scan boundary changes
verifyScanRanges(ptr, scan,
scan.getAttribute(SCAN_ACTUAL_START_ROW),
scan.getStopRow());
}
}
}
```
Basically, we need to take care of these cases:
- Aggregate scan results
- Offset based results
- Dummy cell
- Local index scan boundary changes
- Salt table scan boundary changes
But anyways, fixing group by aggregator cases is required at server side for
data integrity issues. I just wanted to exercise on how many cases we need to
handle here. I am sure local index and salt table scan boundary changes might
also need some careful considerations at server side.
FYI @Divneet18, you can incorporate the changes and write a test with coproc
that can return incorrect result so that we can test scan boundaries negative
results.
> Validate the result returned by phoenix query is within scan range, if
> provided.
> --------------------------------------------------------------------------------
>
> Key: PHOENIX-7004
> URL: https://issues.apache.org/jira/browse/PHOENIX-7004
> Project: Phoenix
> Issue Type: Bug
> Reporter: Rushabh Shah
> Assignee: Divneet Kaur
> Priority: Major
>
> Encountered an issue in production cluster where the range scan query
> returned incorrect results (result were out of range).
> We were running hbck repair operations on the same table. Therefore there
> were region holes and region overlaps exactly at the same time while the
> query was running. Already created PHOENIX-7003 to harden inconsistency
> checks.
> In general, we should validate that the results returned back to customer
> (via PhoenixResultSet#next) is always within the scan boundaries if provided.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
