weizhengte commented on code in PR #8860:
URL: https://github.com/apache/incubator-doris/pull/8860#discussion_r855744904
##########
fe/fe-core/src/main/java/org/apache/doris/statistics/StatisticsJobScheduler.java:
##########
@@ -33,31 +47,211 @@
Switch job state from pending to scheduling.
*/
public class StatisticsJobScheduler extends MasterDaemon {
+ private static final Logger LOG =
LogManager.getLogger(StatisticsJobScheduler.class);
+
+ /**
+ * If the table row-count is greater than the maximum number of Be scans
for a single BE,
+ * we'll divide subtasks by partition. relevant
values(3700000000L&600000000L) are derived from test.
+ * COUNT_MAX_SCAN_PER_TASK is for count(expr), NDV_MAX_SCAN_PER_TASK is
for min(c1)/max(c1)/ndv(c1).
+ */
+ private static final long COUNT_MAX_SCAN_PER_TASK = 3700000000L;
+ private static final long NDV_MAX_SCAN_PER_TASK = 600000000L;
- public Queue<StatisticsJob> pendingJobQueue =
Queues.newLinkedBlockingQueue();
+ /**
+ * Different statistics need to be collected for the jobs submitted by
users.
+ * if all statistics be collected at the same time, the cluster may be
overburdened
+ * and normal query services may be affected. Therefore, we put the jobs
into the queue
+ * and schedule them one by one, and finally divide each job to several
subtasks and execute them.
+ */
+ public final Queue<StatisticsJob> pendingJobQueue =
Queues.newLinkedBlockingQueue(Config.cbo_max_statistics_job_num);
public StatisticsJobScheduler() {
super("Statistics job scheduler", 0);
}
@Override
protected void runAfterCatalogReady() {
- // TODO
- StatisticsJob pendingJob = pendingJobQueue.peek();
- // step0: check job state again
- // step1: divide statistics job to task
- List<StatisticsTask> statisticsTaskList = divide(pendingJob);
- // step2: submit
-
Catalog.getCurrentCatalog().getStatisticsTaskScheduler().addTasks(statisticsTaskList);
+ StatisticsJob pendingJob = this.pendingJobQueue.poll();
+ if (pendingJob != null) {
+ try {
+ List<StatisticsTask> tasks = this.divide(pendingJob);
+
Catalog.getCurrentCatalog().getStatisticsTaskScheduler().addTasks(tasks);
+ pendingJob.updateJobState(StatisticsJob.JobState.SCHEDULING);
+ } catch (DdlException e) {
Review Comment:
ok
--
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]
