You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@hive.apache.org by "anishek (JIRA)" <ji...@apache.org> on 2017/10/17 05:24:00 UTC
[jira] [Updated] (HIVE-17825) Socket not closed when trying to read
files to copy over in replication from metadata
[ https://issues.apache.org/jira/browse/HIVE-17825?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
anishek updated HIVE-17825:
---------------------------
Summary: Socket not closed when trying to read files to copy over in replication from metadata (was: Connection Leak when trying to read files to copy over in replication from metadata)
> Socket not closed when trying to read files to copy over in replication from metadata
> -------------------------------------------------------------------------------------
>
> Key: HIVE-17825
> URL: https://issues.apache.org/jira/browse/HIVE-17825
> Project: Hive
> Issue Type: Bug
> Components: HiveServer2
> Affects Versions: 3.0.0
> Reporter: anishek
> Assignee: anishek
> Priority: Critical
> Fix For: 3.0.0
>
>
> for replication we create a _files in hdfs which lists the source files to be copied over for a table/partition. _files is read in ReplCopyTask to read what files to be copied. The File operations w.r.t to _files is not correct and we leave the files open there, which leads to a lot of CLOSE_WAIT connections to the source Data nodes from HS2 on the replica cluster.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)