[
https://issues.apache.org/jira/browse/SPARK-39866?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Liu Shuo updated SPARK-39866:
-----------------------------
Description:
we are using Spark Thrift Server as a distributed sql query engine, the queries
are related to read some datasource table with a lot of files.
when we open multiple sessions, the Driver could be crashed with OOM.
We can reproduce it with the following steps:
# start spark thrift server;
# using beeline to open a new session;
# create a new datasource table;
# insert one row data into this table;
# open another 5 new sessions, then using `select` command to scan this table
# close all the 6 session
# using jmap command `jmap -histo:live pid > pid.log` to print the dump of
Driver
*Expected result:*
The cached FileStatus should be cleaned, the number of HdfsLocatedFileStatus
object should be 0.
*Actural result:*
** The number of HdfsLocatedFileStatus object is 6.
*!image-2022-07-26-11-54-06-826.png!*
was:
we are using Spark Thrift Server as a distributed sql query engine, the queries
are related to read some datasource table with a lot of files.
when we open multiple sessions, the Driver could be crashed with OOM.
We can reproduce it with the following steps:
# start spark thrift server;
# using beeline to open a new session;
# create a new datasource table;
# insert one row data into this table;
# open another 5 new sessions, then using `select` command to scan this table
# close all the 6 session
# using jmap command to print the dump of Driver
> Memory leak when closing a session of Spark Thrift Server
> ----------------------------------------------------------
>
> Key: SPARK-39866
> URL: https://issues.apache.org/jira/browse/SPARK-39866
> Project: Spark
> Issue Type: Bug
> Components: SQL
> Affects Versions: 3.1.3
> Reporter: Liu Shuo
> Priority: Critical
> Attachments: image-2022-07-26-11-54-06-826.png
>
>
> we are using Spark Thrift Server as a distributed sql query engine, the
> queries are related to read some datasource table with a lot of files.
> when we open multiple sessions, the Driver could be crashed with OOM.
> We can reproduce it with the following steps:
> # start spark thrift server;
> # using beeline to open a new session;
> # create a new datasource table;
> # insert one row data into this table;
> # open another 5 new sessions, then using `select` command to scan this table
> # close all the 6 session
> # using jmap command `jmap -histo:live pid > pid.log` to print the dump of
> Driver
> *Expected result:*
> The cached FileStatus should be cleaned, the number of HdfsLocatedFileStatus
> object should be 0.
> *Actural result:*
> ** The number of HdfsLocatedFileStatus object is 6.
> *!image-2022-07-26-11-54-06-826.png!*
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org
