[
https://issues.apache.org/jira/browse/SPARK-34293?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17276913#comment-17276913
]
Manohar Chamaraju commented on SPARK-34293:
-------------------------------------------
Update:
# In client mode by adding fs.defaultFS in core-site.xml fixed the issue for
me.
# what do to work is usage of hadoop-conf configmap in client mode.
> kubernetes executor pod unable to access secure hdfs
> ----------------------------------------------------
>
> Key: SPARK-34293
> URL: https://issues.apache.org/jira/browse/SPARK-34293
> Project: Spark
> Issue Type: Bug
> Components: Kubernetes
> Affects Versions: 3.0.1
> Reporter: Manohar Chamaraju
> Priority: Major
> Attachments: driver.log, executor.log,
> image-2021-01-30-00-13-18-234.png, image-2021-01-30-00-14-14-329.png,
> image-2021-01-30-00-14-45-335.png, image-2021-01-30-00-20-54-620.png,
> image-2021-01-30-00-33-02-109.png, image-2021-01-30-00-34-05-946.png
>
>
> Steps to reproduce
> # Configure secure HDFS(kerberos) cluster running as containers in
> kubernetes.
> # Configure KDC on centos and create keytab for user principal hdfs, in
> hdfsuser.keytab.
> # Genearte spark image(v3.0.1), to spawn as container out of spark image.
> # Inside spark container, run export HADOOP_CONF_DIR=/etc/hadoop/conf/ with
> core-site.xml configuration as below
> !image-2021-01-30-00-13-18-234.png!
> # Create configmap kbr-conf
> !image-2021-01-30-00-14-14-329.png!
> # Run the command /opt/spark/bin/spark-submit \
> --deploy-mode client \
> --executor-memory 1g\
> --executor-memory 1g\
> --executor-cores 1\
> --class org.apache.spark.examples.HdfsTest \
> --conf spark.kubernetes.namespace=arcsight-installer-lh7fm\
> --master k8s://[https://172.17.17.1:443|https://172.17.17.1/] \
> --conf spark.kubernetes.authenticate.driver.serviceAccountName=spark \
> --conf spark.app.name=spark-hdfs \
> --conf spark.executer.instances=1 \
> --conf spark.kubernetes.node.selector.spark=yes\
> --conf spark.kubernetes.node.selector.Worker=label\
> --conf spark.kubernetes.container.image=manohar/spark:v3.0.1 \
> --conf spark.kubernetes.kerberos.enabled=true \
> --conf spark.kubernetes.kerberos.krb5.configMapName=krb5-conf \
> --conf spark.kerberos.keytab=/data/hdfsuser.keytab \
> --conf spark.kerberos.principal=h...@dom047600.lab \
> local:///opt/spark/examples/jars/spark-examples_2.12-3.0.1.jar \
> hdfs://hdfs-namenode:30820/staging-directory.
> # On running this command driver is able to connect hdfs with kerberos but
> execurtor fails to connect to secure hdfs and below is the logs
> !image-2021-01-30-00-34-05-946.png!
> # Some of observation
> ## In Client mode, --conf spark.kubernetes.hadoop.configMapName=hadoop-conf
> as not effect only works after HADOOP_CONF_DIR is set. Below was the contents
> of hadoop-conf configmap.
> !image-2021-01-30-00-20-54-620.png!
> ## Ran the command in cluster mode as well, in cluster mode also executor
> could not connect to secure hdfs.
>
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org
