You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@sentry.apache.org by "Hadoop QA (JIRA)" <ji...@apache.org> on 2017/01/21 05:29:27 UTC
[jira] [Commented] (SENTRY-1605) SENTRY-1508 need to be fixed
because of Kerberos initialization issue
[ https://issues.apache.org/jira/browse/SENTRY-1605?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15832819#comment-15832819 ]
Hadoop QA commented on SENTRY-1605:
-----------------------------------
Here are the results of testing the latest attachment
https://issues.apache.org/jira/secure/attachment/12848685/SENTRY-1605.patch against master.
{color:green}Overall:{color} +1 all checks pass
{color:green}SUCCESS:{color} all tests passed
Console output: https://builds.apache.org/job/PreCommit-SENTRY-Build/2271/console
This message is automatically generated.
> SENTRY-1508 need to be fixed because of Kerberos initialization issue
> ---------------------------------------------------------------------
>
> Key: SENTRY-1605
> URL: https://issues.apache.org/jira/browse/SENTRY-1605
> Project: Sentry
> Issue Type: Bug
> Reporter: Vadim Spector
> Assignee: Vadim Spector
> Attachments: SENTRY-1605.patch
>
>
> SENTRY-1508 performs full Sentry HDFS sync at the construction time. Turns out, it does not work with Kerberos on, because the Hive code initialized Kerberos client only AFTER instantiating MetastorePlugin. Fix is pretty straightforward: MetastorePlugin should not attempt synchronization with Sentry from the constructor. It should rely on the housekeeping thread's SyncTask to ultimately succeed in pushing the first full update. The subtle point is that until it happens, all partial updates going through MetastorePlugin should skip Sentry notification step while keeping updating the MetastorePlugin's local cache as usual.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)