Sahara

Spark plugin can't use swift as output datasource

Bug #1524997 reported by Vitalii Gridnev
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
Sahara
Fix Released
High
Vitalii Gridnev
Sahara mitaka-2 "m2"
Liberty
Fix Released
High
Vitalii Gridnev

Bug Description

Spark plugin can't use swift as output datasource, because of the problems with permissions during creating files in hadoop.tmp.dir

Log of the job execution:

Exception in thread "main" org.apache.spark.SparkException: Job aborted due to stage failure: Task 0 in stage 2.0 failed 4 times, most recent failure: Lost task 0.3 in stage 2.0 (TID 7, spark-cluster-spark-worker-001.novalocal): java.io.IOException: Permission denied

Tags: plugin.spark
Revision history for this message
OpenStack Infra (hudson-openstack) wrote : Fix proposed to sahara (master)

Fix proposed to branch: master
Review: https://review.openstack.org/256317

Changed in sahara:
status: Triaged → In Progress
Revision history for this message
OpenStack Infra (hudson-openstack) wrote : Fix merged to sahara (master)

Reviewed: https://review.openstack.org/256317
Committed: https://git.openstack.org/cgit/openstack/sahara/commit/?id=3924d3e3bf8261424386e513d545dd610c28d63f
Submitter: Jenkins
Branch: master

commit 3924d3e3bf8261424386e513d545dd610c28d63f
Author: Vitaly Gridnev <email address hidden>
Date: Fri Dec 11 13:08:00 2015 +0300

    Don't configure hadoop.tmp.dir in Spark plugin

    During job execution with Swift spark-submit writes some data in
    hadoop.tmp.dir. But for executor user there is no permissions for
    writing in current directory. So, it's suggested to use default value
    for this config options as it's done for all other plugins.

    Change-Id: I0e2ec932cfd4fc49cd0f20098badb15c18769e20
    Closes-bug: 1524997

Changed in sahara:
status: In Progress → Fix Released
Revision history for this message
OpenStack Infra (hudson-openstack) wrote : Fix proposed to sahara (stable/liberty)

Fix proposed to branch: stable/liberty
Review: https://review.openstack.org/259977

Revision history for this message
OpenStack Infra (hudson-openstack) wrote : Fix merged to sahara (stable/liberty)

Reviewed: https://review.openstack.org/259977
Committed: https://git.openstack.org/cgit/openstack/sahara/commit/?id=c716f5b723045ab7651b8ac3214f4ef046ba26c8
Submitter: Jenkins
Branch: stable/liberty

commit c716f5b723045ab7651b8ac3214f4ef046ba26c8
Author: Vitaly Gridnev <email address hidden>
Date: Fri Dec 11 13:08:00 2015 +0300

    Don't configure hadoop.tmp.dir in Spark plugin

    During job execution with Swift spark-submit writes some data in
    hadoop.tmp.dir. But for executor user there is no permissions for
    writing in current directory. So, it's suggested to use default value
    for this config options as it's done for all other plugins.

    Change-Id: I0e2ec932cfd4fc49cd0f20098badb15c18769e20
    Closes-bug: 1524997
    (cherry picked from commit 3924d3e3bf8261424386e513d545dd610c28d63f)

To post a comment you must log in.
This report contains Public information  
Everyone can see this information.
Subscribing...

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.