Mailing-List: contact yarn-issues-help@hadoop.apache.org; run by ezmlm
Precedence: bulk
Date: Wed, 1 Jun 2016 23:37:59 +0000 (UTC)
From: "Vrushali C (JIRA)" <jira@apache.org>
To: yarn-issues@hadoop.apache.org
Message-ID: <JIRA.12972887.1464213986000.8921.1464824279410@Atlassian.JIRA>
In-Reply-To: <JIRA.12972887.1464213986000@Atlassian.JIRA>
References: <JIRA.12972887.1464213986000@Atlassian.JIRA> <JIRA.12972887.1464213986141@arcas>
Subject: [jira] [Commented] (YARN-5156) YARN_CONTAINER_FINISHED of
 YARN_CONTAINERs will always have running state
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit
archived-at: Wed, 01 Jun 2016 23:38:01 -0000


    [ https://issues.apache.org/jira/browse/YARN-5156?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15311369#comment-15311369 ] 

Vrushali C commented on YARN-5156:
----------------------------------


While the change in NM may be broader, I think we have two options at our end:
- not store the final status in the container finished event
- store it as complete if it's incorrectly set.

For #1, it sort of leads to inconsistent contents across events. With hRaven, the experience has been that applications that used hRaven like reducer estimation would query for "finished" or "completed" jobs, and if such fields were missing, it lead to incorrect calculations at their end. For example, in hRaven I patched a very similar issue : the jobStatus field seemed to occur only in JOB_INITED, JOB_KILLED AND JOB_FAILED events. It was missing from the JOB_FINISHED event. The jobStatus field should be a part of the JOB_FINISHED event in the history file when it's generated.
https://github.com/twitter/hraven/issues/72

So my vote is for at least ensuring we store some consistent information for all events. In case of container finished, it is reasonable to think that the container had completed. I am attaching a very basic patch. I am still checking if that helps fix this issue in my pseudo-distributed cluster. 


> YARN_CONTAINER_FINISHED of YARN_CONTAINERs will always have running state
> -------------------------------------------------------------------------
>
>                 Key: YARN-5156
>                 URL: https://issues.apache.org/jira/browse/YARN-5156
>             Project: Hadoop YARN
>          Issue Type: Sub-task
>          Components: timelineserver
>            Reporter: Li Lu
>         Attachments: YARN-5156-YARN-2928.01.patch
>
>
> On container finished, we're reporting "YARN_CONTAINER_STATE: "RUNNING"". Do we design this deliberately or it's a bug? 
> {code}
> {
> metrics: [ ],
> events: [
> {
> id: "YARN_CONTAINER_FINISHED",
> timestamp: 1464213765890,
> info: {
> YARN_CONTAINER_EXIT_STATUS: 0,
> YARN_CONTAINER_STATE: "RUNNING",
> YARN_CONTAINER_DIAGNOSTICS_INFO: ""
> }
> },
> {
> id: "YARN_NM_CONTAINER_LOCALIZATION_FINISHED",
> timestamp: 1464213761133,
> info: { }
> },
> {
> id: "YARN_CONTAINER_CREATED",
> timestamp: 1464213761132,
> info: { }
> },
> {
> id: "YARN_NM_CONTAINER_LOCALIZATION_STARTED",
> timestamp: 1464213761132,
> info: { }
> }
> ],
> id: "container_e15_1464213707405_0001_01_000018",
> type: "YARN_CONTAINER",
> createdtime: 1464213761132,
> info: {
> YARN_CONTAINER_ALLOCATED_PRIORITY: "20",
> YARN_CONTAINER_ALLOCATED_VCORE: 1,
> YARN_CONTAINER_ALLOCATED_HOST_HTTP_ADDRESS: "10.22.16.164:0",
> UID: "yarn_cluster!application_1464213707405_0001!YARN_CONTAINER!container_e15_1464213707405_0001_01_000018",
> YARN_CONTAINER_ALLOCATED_HOST: "10.22.16.164",
> YARN_CONTAINER_ALLOCATED_MEMORY: 1024,
> SYSTEM_INFO_PARENT_ENTITY: {
> type: "YARN_APPLICATION_ATTEMPT",
> id: "appattempt_1464213707405_0001_000001"
> },
> YARN_CONTAINER_ALLOCATED_PORT: 64694
> },
> configs: { },
> isrelatedto: { },
> relatesto: { }
> }
> {code}


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

---------------------------------------------------------------------
To unsubscribe, e-mail: yarn-issues-unsubscribe@hadoop.apache.org
For additional commands, e-mail: yarn-issues-help@hadoop.apache.org