-
Notifications
You must be signed in to change notification settings - Fork 34
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Possible issue with --packages argument for submit-args #90
Comments
Using the --packages argument seems to actually populate the --jars argument with local paths to all .jar dependencies, and that won't work inside the cluster. What I ended up doing is a combination of the
|
Hi, I had the same issue with packages. spark-submit \
--packages org.mongodb.spark:mongo-spark-connector_2.10:2.0.0 \
count_something.py But to be able run it with DC/OS Spark I have to do the follow: dcos spark run \
--submit-args=""`
`"--conf spark.mesos.uris=http://repo1.maven.org/maven2/org/mongodb/spark/mongo-spark-connector_2.10/2.0.0/mongo-spark-connector_2.10-2.0.0.jar,http://repo1.maven.org/maven2/
org/mongodb/mongo-java-driver/3.2.2/mongo-java-driver-3.2.2.jar "`
`"--conf spark.driver.extraClassPath=/mnt/mesos/sandbox/mongo-spark-connector_2.10-2.0.0.jar:/mnt/mesos/sandbox/mongo-java-driver-3.2.2.jar "`
`"--conf spark.executor.extraClassPath=/mnt/mesos/sandbox/mongo-spark-connector_2.10-2.0.0.jar:/mnt/mesos/sandbox/mongo-java-driver-3.2.2.jar "`
`"https://example.com/spark/do_something.py" \
--verbose As you can see it cannot resolve dependicies and this why I had to provide also jar for I think it is only because dcos spark bypasses |
@bogdangi |
@andreimarinescu there is a fix here for driver dependencies: apache/spark#18587. It will go in 2.3. Also planning to update the cli, --packages is currently unsupported along with some other flags. |
Hello everyone,
I'm trying to run a job using the following command:
The task fails immediately with the following output from dcos spark status <job_id>:
Looking at the mesos logs on that agent, I'm getting the following output. As far as I can see, all the jar dependencies are downloaded locally and they're only referenced from the mesos agent. Indeed, if I delete everything in /home/andrei/.ivy2, submitting the job takes substantially longer and the folder is then repopulated. Is this a bug, or am I doing anything wrong? I haven't found anything in the docs so far regarding the --packages parameter, running this on my local Spark cluster works.
The text was updated successfully, but these errors were encountered: