Mailing-List: contact common-issues-help@hadoop.apache.org; run by ezmlm
Precedence: bulk
Reply-To: common-issues@hadoop.apache.org
Date: Mon, 29 Sep 2014 17:55:35 +0000 (UTC)
From: "Andrew Wang (JIRA)" <jira@apache.org>
To: common-issues@hadoop.apache.org
Message-ID: <JIRA.12742931.1411194052000.146802.1412013335129@Atlassian.JIRA>
In-Reply-To: <JIRA.12742931.1411194052000@Atlassian.JIRA>
References: <JIRA.12742931.1411194052000@Atlassian.JIRA>
 <JIRA.12742931.1411194052155@arcas>
Subject: [jira] [Commented] (HADOOP-11113) Namenode not able to reconnect to
 KMS after KMS restart
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit


    [ https://issues.apache.org/jira/browse/HADOOP-11113?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14151957#comment-14151957 ] 

Andrew Wang commented on HADOOP-11113:
--------------------------------------

Thanks for working on this Arun, just a few review comments:

* Re: status codes, 401 requires that the WWW-Authenticate header is set. I think we're stuck without an appropriate error code here, so we went with 403. [~tucu00] might remember this better than me though.
* Commented out kdcConf lines in TestKMS could be removed
* Rather than modify all the runServer lines, you could do some method delegation (have one version without a port param, have it call the one with a port param).

> Namenode not able to reconnect to KMS after KMS restart
> -------------------------------------------------------
>
>                 Key: HADOOP-11113
>                 URL: https://issues.apache.org/jira/browse/HADOOP-11113
>             Project: Hadoop Common
>          Issue Type: Bug
>          Components: security
>            Reporter: Arun Suresh
>            Assignee: Arun Suresh
>         Attachments: HADOOP-11113.1.patch
>
>
> It is observed that if KMS is restarted without the Namenode being restarted, NN will not be able to reconnect with KMS.
> It seems that the KMS auth cookie goes stale and it does not get flushed, so the KMSClient in the NN cannot reconnect with the new KMS.


--
This message was sent by Atlassian JIRA
(v6.3.4#6332)