github-actions[bot] commented on code in PR #63625:
URL: https://github.com/apache/doris/pull/63625#discussion_r3302327693
##########
fe/fe-core/src/main/java/org/apache/doris/statistics/OlapAnalysisTask.java:
##########
@@ -330,14 +330,22 @@ protected void doFull() throws Exception {
doPartitionTable();
Review Comment:
This path bypasses explicit hot-value collection for partitioned OLAP
tables. When `enable_partition_analyze` is on and the table is partitioned,
`doFull()` calls `doPartitionTable()` before checking
`shouldCollectHotValue()`, and the partition merge SQL
(`MERGE_PARTITION_TEMPLATE`) always emits `null as hot_value`. That means
`ANALYZE TABLE ... WITH HOT VALUE` silently produces no hot values for
partitioned tables. Please either route explicit hot-value full analyze through
the hot-value SQL path or make the partition merge path collect/merge hot
values, and add partitioned coverage.
##########
fe/fe-core/src/main/java/org/apache/doris/statistics/AnalysisManager.java:
##########
@@ -381,6 +381,9 @@ public AnalysisInfo
buildAnalysisJobInfo(AnalyzeTableCommand command) {
infoBuilder.setCronExpression(cronExpression);
infoBuilder.setForceFull(command.forceFull());
infoBuilder.setUsingSqlForExternalTable(command.usingSqlForExternalTable());
+ AnalyzeProperties analyzeProperties = command.getAnalyzeProperties();
+ infoBuilder.setCollectHotValue((analyzeProperties.hasCollectHotValue()
Review Comment:
This new default is not honored by the external-table full analyze path.
`buildAnalysisJobInfo()` now sets `collectHotValue=false` for default full
analyze, but `ExternalAnalysisTask.doFull()` ignores `info.collectHotValue`,
always fills the hot-value params, and always uses `FULL_ANALYZE_TEMPLATE`. As
a result, external full analyze still performs the memory-heavy hot-value
aggregation by default, which leaves the reported problem unfixed for that
parallel path. Please gate `ExternalAnalysisTask.doFull()` with the same
flag/template choice and add coverage for external full analyze.
--
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]
