github-actions[bot] commented on code in PR #68282:
URL: https://github.com/apache/doris/pull/68282#discussion_r4256994607
##########
fe/fe-core/src/main/java/org/apache/doris/statistics/analysis/TableStatsMeta.java:
##########
@@ -130,6 +140,94 @@ public TableStatsMeta(long rowCount, AnalysisInfo
analyzedJob, TableIf table) {
update(analyzedJob, table);
}
+ /**
+ * Create a record for a table which doesn't have one yet, in the state of
an empty table. The rows
+ * loaded into the table are accumulated by {@link
AnalysisManager#replayUpdateRowsRecord}, so a record
+ * has to exist before the first load, otherwise these rows can never be
turned into a row count.
+ */
+ public TableStatsMeta(OlapTable table) {
Review Comment:
[P3] Keep empty records from marking tables as analyzed
StatisticsMetricCollector increments statistics_not_analyzed_table_num only
when findTableStatsStatus(tableId) is null (lines 89-91). A nonempty table that
has never been analyzed therefore stops contributing to that gauge as soon as
TRUNCATE creates this empty record, even though it still has no column stats or
analysis timestamp. Update the metric's consumer to distinguish this retained
empty state from analyzed or user-injected statistics; otherwise the dashboard
undercounts tables needing their first analysis.
##########
fe/fe-core/src/main/java/org/apache/doris/nereids/trees/plans/commands/ShowTableStatsCommand.java:
##########
@@ -242,28 +242,37 @@ public ShowResultSet
constructTableResultSet(TableStatsMeta tableStatistic, Tabl
row.add(String.valueOf(tableStatistic.updatedRows));
row.add(String.valueOf(tableStatistic.queriedTimes.get()));
row.add(String.valueOf(tableStatistic.rowCount));
- LocalDateTime dateTime =
-
LocalDateTime.ofInstant(Instant.ofEpochMilli(tableStatistic.updatedTime),
- java.time.ZoneId.systemDefault());
- LocalDateTime lastAnalyzeTime =
-
LocalDateTime.ofInstant(Instant.ofEpochMilli(tableStatistic.lastAnalyzeTime),
- java.time.ZoneId.systemDefault());
- row.add(dateTime.format(formatter));
+ row.add(formatStatsTime(tableStatistic.updatedTime, formatter));
Set<Pair<String, String>> columnsSet = tableStatistic.analyzeColumns();
Set<Pair<String, String>> newColumnsSet = new HashSet<>();
for (Pair<String, String> pair : columnsSet) {
newColumnsSet.add(Pair.of(Util.getTempTableDisplayName(pair.first),
pair.second));
}
row.add(newColumnsSet.toString());
- row.add(tableStatistic.jobType.toString());
+ // A record which was not created by an analyze job, e.g. the record
kept by TRUNCATE TABLE, has no
+ // trigger to show.
+ row.add(tableStatistic.jobType == null ? "" :
tableStatistic.jobType.toString());
Review Comment:
[P2] Read the nullable trigger once when showing table stats
SHOW TABLE STATS holds neither the table lock nor this record's monitor. On
a previously analyzed table, the first jobType read here can be non-null, then
a concurrent TRUNCATE TABLE runs TableStatsMeta.reset() and sets the same
retained record's jobType to null before the second read calls toString(). The
SHOW statement then fails with NullPointerException despite the null guard.
Capture jobType once in a local variable, or read a snapshot under the record
monitor.
--
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]
