skonto commented on a change in pull request #23546: [SPARK-23153][K8s] Support
client dependencies with a Hadoop Compatible File System
URL: https://github.com/apache/spark/pull/23546#discussion_r259157717
##########
File path: core/src/main/scala/org/apache/spark/deploy/SparkSubmit.scala
##########
@@ -318,9 +319,25 @@ private[spark] class SparkSubmit extends Logging {
args.ivySettingsPath)
if (!StringUtils.isBlank(resolvedMavenCoordinates)) {
- args.jars = mergeFileLists(args.jars, resolvedMavenCoordinates)
- if (args.isPython || isInternal(args.primaryResource)) {
- args.pyFiles = mergeFileLists(args.pyFiles, resolvedMavenCoordinates)
+ // In K8s client mode, when in the driver, add resolved jars early as
we might need
+ // them at the submit time. For example we might use the dependencies
for downloading
+ // files from a Hadoop Compatible fs eg. S3. In this case the user
might pass:
+ // --packages
com.amazonaws:aws-java-sdk:1.7.4:org.apache.hadoop:hadoop-aws:2.7.6
+ if (isKubernetesClient &&
+ args.sparkProperties.getOrElse("spark.kubernetes.submitInDriver",
"false").toBoolean) {
Review comment:
it means its the first submit time in cluster mode. So at this step the
submit process needs to upload the dependencies. In practice this is user's
machine. This is where I also need ot make sure packages works because user may
need to pass the dependencies needed for accessing the remote system. There are
a bunch of places this needs to work. So I just make the same packages
available at the user's machine and in the driver because we walso want to pass
the deps in the there.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
With regards,
Apache Git Services
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
