github-actions[bot] commented on code in PR #66756:
URL: https://github.com/apache/doris/pull/66756#discussion_r3791889790
##########
fe/fe-core/src/main/java/org/apache/doris/statistics/ColStatsData.java:
##########
@@ -139,6 +139,10 @@ public String toSQL(boolean roundByParentheses) {
public ColumnStatistic toColumnStatistic() {
try {
+ if (!isValid()) {
Review Comment:
[P2] Avoid overflow in the newly activated ratio check — This guard now
applies `isValid()` to manual cache publication and persisted-row reloads, but
its `ndv > 10 * count` (and later `nullCount * 10`) arithmetic is signed
`long`. A valid supported row such as `row_count=1000000000000000000, ndv=1`
makes `10 * count` wrap negative, so both the immediate cache and reload become
UNKNOWN and SHOW hides the row; the internal statistics fields are BIGINT and
ALTER STATS accepts these numeric inputs. Please express both factor-of-ten
comparisons without integral multiplication overflow and cover values above
`Long.MAX_VALUE / 10` through both conversion paths.
##########
fe/fe-core/src/main/java/org/apache/doris/statistics/ColumnStatistic.java:
##########
@@ -142,6 +142,11 @@ public static ColumnStatistic
fromResultRowList(List<ResultRow> resultRows) {
* this function is used by analyze job and cbo job.
*/
public static ColumnStatistic fromResultRow(ResultRow row) {
+ ColStatsData statsData = new ColStatsData(row);
+ if (!statsData.isValid()) {
Review Comment:
[P2] Keep retained invalid rows from flooding WARN on every read —
`isValid()` logs on every false branch, so this new reader guard makes the
intentionally persisted UNKNOWN representation noisy at every boundary:
collection already validates and logs, `syncColStats` validates again, each
follower validates again, and preheat/reload/refresh/SHOW repeat it for the
retained row. Preheat alone can examine up to the default 500,000 cache
entries. Please make the reader-side check quiet (or make validity testing
side-effect-free) and emit the warning/metric once at the collection boundary.
--
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]
