[
https://issues.apache.org/jira/browse/SPARK-11356?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14997651#comment-14997651
]
Xin Wu commented on SPARK-11356:
--------------------------------
[~maver1ck]
If I only execute {code}sqlContext.sql("select * from t where day =
20151027").count(){code} everytime after a new append, I don't get updated
count.. but If I run
{code}
sqlContext.read.parquet("some_location").registerTempTable("t")
sqlContext.sql("select * from t where day = 20151027").count()
{code}
Then, the count result is updated according to the newly appended records.
Do you expect to see updated count only running the select statement?
> Option to refresh information about parquet partitions
> ------------------------------------------------------
>
> Key: SPARK-11356
> URL: https://issues.apache.org/jira/browse/SPARK-11356
> Project: Spark
> Issue Type: Improvement
> Components: SQL
> Affects Versions: 1.5.1
> Reporter: Maciej Bryński
>
> I have two apps:
> 1) First one periodically append data to parquet (which creates new partition)
> 2) Second one executes query on data
> Right now I can't find any possibility to force Spark to make partition
> discovery. So every query is executed on the same data.
> I tried --conf spark.sql.parquet.cacheMetadata=false but without success.
> Is there any option to make this happen ?
> App 1 - periodically (eg. every hour)
> {code}
> df.write.partitionBy("day").mode("append").parquet("some_location")
> {code}
> App 2 - example
> {code}
> sqlContext.read.parquet("some_location").registerTempTable("t")
> sqlContext.sql("select * from t where day = 20151027").count()
> {code}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org
