You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@hbase.apache.org by "mazhenlin (JIRA)" <ji...@apache.org> on 2018/10/18 08:27:00 UTC
[jira] [Commented] (HBASE-15291) FileSystem not closed in secure
bulkLoad
[ https://issues.apache.org/jira/browse/HBASE-15291?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16654855#comment-16654855 ]
mazhenlin commented on HBASE-15291:
-----------------------------------
[~ashish singhi]
As far as i can see ,the race condition mentioned here still exists, even after using FileSystem#newInstance. That is because FileSystems are closed by calling FileSystem.closeAllForUGI(ugi) int the cleanupBulkLoad function, and FileSystems created by FileSystem#newInstance do share the same user, so they will be closed by the FileSystem.closeAllForUGI(ugi) call.
Solution to this race condition might be more complicated ,e.g. reference counting
> FileSystem not closed in secure bulkLoad
> ----------------------------------------
>
> Key: HBASE-15291
> URL: https://issues.apache.org/jira/browse/HBASE-15291
> Project: HBase
> Issue Type: Bug
> Affects Versions: 1.0.2, 0.98.16.1
> Reporter: Yong Zhang
> Assignee: Ashish Singhi
> Priority: Major
> Fix For: 3.0.0, 2.1.0, 1.5.0, 1.3.3, 1.4.4, 2.0.1, 1.2.7
>
> Attachments: HBASE-15291-branch-1.2.patch, HBASE-15291-branch-1.patch, HBASE-15291-revert-master.patch, HBASE-15291.001.patch, HBASE-15291.002.patch, HBASE-15291.003.patch, HBASE-15291.004.patch, HBASE-15291.addendum, HBASE-15291.patch, HBASE-15291.v1.patch, HBASE-15291.v2.patch, HBASE-15291.v2.patch, patch
>
>
> FileSystem not closed in secure bulkLoad after bulkLoad finish, it will cause memory used more and more if too many bulkLoad .
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)