deshanxiao commented on a change in pull request #24102: [SPARK-27171][SQL]
Support Full-Partitons scan in limit for the first time
URL: https://github.com/apache/spark/pull/24102#discussion_r274864400
##########
File path:
sql/core/src/main/scala/org/apache/spark/sql/execution/SparkPlan.scala
##########
@@ -344,7 +344,7 @@ abstract class SparkPlan extends QueryPlan[SparkPlan] with
Logging with Serializ
while (buf.size < n && partsScanned < totalParts) {
// The number of partitions to try in this iteration. It is ok for this
number to be
// greater than totalParts because we actually cap it at totalParts in
runJob.
- var numPartsToTry = 1L
+ var numPartsToTry =
Math.max(Math.ceil(sqlContext.conf.limitStartUpFactor * totalParts).toLong, 1L)
Review comment:
@cloud-fan Sorry for so late reply. I try to execute a very simple query:
```
select * from tst where data='20180410' and name like 'xx' limit 1; (400
partitions)
```
It creates many jobs because "limit" will scan more partitions as the retry
increasing (1 -> 4 -> 20 -> 100 -> 200). When I try to add this parameter
(0.1). The scan partition from 40 to 200. I think the parameter will increase
parallelism to speed up the query. So, different queries may improve
performance differently.
----------------------------------------------------------------
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.
For queries about this service, please contact Infrastructure at:
[email protected]
With regards,
Apache Git Services
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
