Mailing-List: contact yarn-dev-help@hadoop.apache.org; run by ezmlm
Precedence: bulk
Reply-To: yarn-dev@hadoop.apache.org
Date: Tue, 2 Jun 2015 10:23:18 +0000 (UTC)
From: "Steve Loughran (JIRA)" <jira@apache.org>
To: yarn-dev@hadoop.apache.org
Message-ID: <JIRA.12834545.1433240582000.92495.1433240598854@Atlassian.JIRA>
In-Reply-To: <JIRA.12834545.1433240582000@Atlassian.JIRA>
References: <JIRA.12834545.1433240582000@Atlassian.JIRA>
 <JIRA.12834545.1433240582454@arcas>
Subject: [jira] [Created] (YARN-3759) Include command line, localization
 info and env vars on AM launch failure
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: quoted-printable

Steve Loughran created YARN-3759:
------------------------------------

             Summary: Include command line, localization info and env vars =
on AM launch failure
                 Key: YARN-3759
                 URL: https://issues.apache.org/jira/browse/YARN-3759
             Project: Hadoop YARN
          Issue Type: Sub-task
          Components: nodemanager
    Affects Versions: 2.7.0
            Reporter: Steve Loughran
            Priority: Minor


While trying to diagnose AM launch failures, its important to be able to ge=
t at the final, expanded {{CLASSPATH}} and other env variables. We don't ge=
t that today: you can log the unexpanded values on the client, and tweak NM=
 ContainerExecutor log levels to DEBUG & get some of this =E2=80=94=E2=80=
=9Aut you don't get it in the task logs, and tuning NM log level isn't viab=
le on a large, busy cluster.

Launch failures should include some env specifics:
# list of env vars (ideally, full getenv values), with some stripping of "s=
ensitive" options (i'm thinking AWS env vars here)
# command line
# path localisations

These can go in the task logs, we don't need to include them in the applica=
tion report.


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