yujun777 commented on code in PR #69896:
URL: https://github.com/apache/doris/pull/69896#discussion_r4462608419
##########
fe/fe-core/src/main/java/org/apache/doris/statistics/analysis/TableStatsMeta.java:
##########
@@ -391,6 +397,12 @@ public long getBaseIndexDeltaRowCount(OlapTable table) {
* {@link #updatedRows} and reports them as delta rows.
*/
public synchronized long getRowCountWithDeltaRows(OlapTable table, long
indexId) {
Review Comment:
The StatsCalculator shortcut for injected statistics is a consumer-side
issue to address with explicit wrapper-estimate priority in #69850; this PR
does not change that shortcut.
In 9fb10c4c0e1 the separate TableStatsMeta accumulator was removed.
Changelog deltas now update the partition MaterializedIndex row count, which
the P/Q metadata lookup reads independently of injected base statistics. This
supplies the source count, while wrapper estimate selection remains in the
consumer PR.
##########
fe/fe-core/src/main/java/org/apache/doris/statistics/analysis/TableStatsMeta.java:
##########
@@ -414,6 +426,31 @@ public synchronized long
getRowCountWithDeltaRows(OlapTable table, long indexId)
return rowCount + getBaseIndexDeltaRowCount(table);
}
+ /**
+ * Add the rows a load wrote to an index which has no collected row count
of its own.
+ */
+ public synchronized void addIndexUpdatedRows(long indexId, long rows) {
+ if (rows <= 0) {
+ return;
Review Comment:
Fixed in 9fb10c4c0e1 by removing the lifetime accumulator. Loads now adjust
the partition's changelog MaterializedIndex estimate, and the existing periodic
tablet-stat path replaces that estimate with the retained row count, including
TTL removal. The estimate can still be stale until a fresh report arrives; it
no longer permanently accumulates expired rows in a separate persisted map.
The new test replaces the estimate with a lower reported retained count and
verifies that the next load is added to that refreshed baseline.
##########
fe/fe-core/src/main/java/org/apache/doris/statistics/analysis/AnalysisManager.java:
##########
@@ -1525,6 +1535,7 @@ public void replayUpdateRowsRecord(UpdateRowsEvent event)
{
+ "rows for all replica {}, tablets count {}",
olapTable.getName(), tableUpdateRows, replicaNum,
rowsForAllReplica, tabletRows.size());
Review Comment:
Fixed in 9fb10c4c0e1. Each changelog tablet's aggregate delta is divided by
its owning partition's replica allocation before updating that partition's
index estimate. It is no longer divided by the table default.
The new test uses a table default of three replicas with one- and
three-replica partitions: reported totals of 9 and 18 add 9 and 6 rows
respectively, while changelog rows remain excluded from the base counter.
--
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]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
