I've been playing around with HDFS and Spark. I've set up a five node cluster on my network running HDFS, Spark, and managed by Yarn. Workers are running in client mode.
From the master node, I can launch the PySpark shell just fine. Running example jars, the job is split up to the worker nodes and executes nicely.
I have a few questions on whether and how to run python/Pyspark files against this cluster.
If I have a python file with a PySpark calls elsewhere else, like on my local dev laptop or a docker container somewhere, is there a way to run or submit this file locally and have it executed on the remote Spark cluster? Methods that I'm wondering about involve running spark-submit in the local/docker environment and but the file has SparkSession.builder.master() configured to the remote cluster.
Related, I see a configuration for --master in spark-submit, but the only yarn option is to pass "yarn" which seems to only queue locally? Is there a way to specify remote yarn?
If I can set up and run the file remotely, how do I set up SparkSession.builder.master()? Is the url just to the hdfs:// url to port 9000, or do I submit it to one of the Yarn ports?
TIA!
way to run or submit this file locally and have it executed on the remote Spark cluster
Yes, well "YARN", not "remote Spark cluster". You set --master=yarn when running with spark-submit, and this will run against the configured yarn-site.xml in HADOOP_CONF_DIR environment variable. You can define this at the OS level, or in spark-env.sh.
You can also use SparkSession.builder.master('yarn') in code. If both options are supplied, one will get overridden.
To run fully "in the cluster", also set --deploy-mode=cluster
Is there a way to specify remote yarn?
As mentioned, this is configured from yarn-site.xml for providing resourcemanager location(s).
how do I set up SparkSession.builder.master()? Is the url just to the hdfs:// url to port 9000
No - The YARN resource manager has its own RPC protocol, not hdfs:// ... You can use spark.read("hdfs://namenode:port/path") to read HDFS files, though. As mentioned, .master('yarn') or --master yarn are the only configs you need that are specific for Spark.
If you want to use Docker containers, YARN does support this, but Spark's Kubernetes master will be easier to setup, and you can use Hadoop Ozone or MinIO rather than HDFS in Kubernetes.
Related
I'm running spark on standalone mode as a docker service where I have one master node and one spark worker. I followed the spark documentation instructions:
https://spark.apache.org/docs/latest/spark-standalone.html
to add the properties where the spark cluster cleans itself and I set those in my docker_entrypoint
export SPARK_WORKER_OPTS="-Dspark.worker.cleanup.enabled=true -Dspark.worker.cleanup.interval=900 -Dspark.worker.cleanup.appDataTtl=900
and verify that it was enables following the logs of the worker node service
My question is do we expect to get all directories located on SPARK_WORKER_DIR directory to be cleaned ? or does it only clean the application files
Because I still see some empty directories holding there
I have a cluster with two workers and one master.
To start master & workers I use the sbin/start-master.sh and sbin/start-slaves.shin the master's machine. Then, the master UI shows me that the slaves are ALIVE (so, everything OK so far). Issue comes when I want to use spark-submit.
I execute this command in my local machine:
spark-submit --master spark://<master-ip>:7077 --deploy-mode cluster /home/user/example.jar
But the following error pops up: ERROR ClientEndpoint: Exception from cluster was: java.nio.file.NoSuchFileException: /home/user/example.jar
I have been doing some research in stack overflow and Spark's documentation and it seems like I should specify the application-jar of spark-submit command as "Path to a bundled jar including your application and all dependencies. The URL must be globally visible inside of your cluster, for instance, an hdfs:// path or a file:// path that is present on all nodes." (as it indicates https://spark.apache.org/docs/latest/submitting-applications.html).
My question is: how can I set my .jar as globally visible inside the cluster? There is a similar question in here Spark Standalone cluster cannot read the files in local filesystem but solutions do not work for me.
Also, am I doing something wrong by initialising the cluster inside my master's machine using sbin/start-master.sh but then doing the spark-submit in my local machine? I initialise the master inside my master's terminal because I read so in Spark's documentation, but maybe this has something to do with the issue. From Spark's documentation:
Once you’ve set up this file, you can launch or stop your cluster with the following shell scripts, based on Hadoop’s deploy scripts, and available in SPARK_HOME/sbin: [...] Note that these scripts must be executed on the machine you want to run the Spark master on, not your local machine.
Thank you very much
EDIT:
I have copied the file .jar in every worker and it works. But my point is to know if there is a better way, since this method makes me copy the .jar to each worker everytime I create a new jar. (This was one of the answers from the question of the already posted link Spark Standalone cluster cannot read the files in local filesystem )
#meisan your spark-submit command is missing out on 2 things.
your jars should be added with flag --jar
file holding your driver code i.e. the main function.
Now you have not specified anywhere if you are using scala or python but in the nutshell your command will look something like:
for python :
spark-submit --master spark://<master>:7077 --deploy-mode cluster --jar <dependency-jars> <python-file-holding-driver-logic>
for scala:
spark-submit --master spark://<master>:7077 --deploy-mode cluster --class <scala-driver-class> --driver-class-path <application-jar> --jar <dependency-jars>
Also, spark takes care of sending the required files and jars to the executors when you use the documented flags.
If you want to omit the --driver-class-path flag, you can set the environmental variable SPARK_CLASSPATH to path where all your jars are placed.
We use python with pyspark api in order to run simple code on spark cluster.
from pyspark import SparkContext, SparkConf
conf = SparkConf().setAppName('appName').setMaster('spark://clusterip:7077')
sc = SparkContext(conf=conf)
rdd = sc.parallelize([1, 2, 3, 4])
rdd.map(lambda x: x**2).collect()
It works when we setup a spark cluster locally and with dockers.
We would now like to start an emr cluster and test the same code. And seems that pyspark can't connect to the spark cluster on emr
We opened ports 8080 and 7077 from our machine to the spark master
We are getting past the firewall and just seems that nothing is listening on port 7077 and we get connection refused.
We found this explaining how to serve a job using the cli but we need to run it directly from pyspark api on the driver.
What are we missing here?
How can one start an emr cluster and actually run pyspark code locally on python using this cluster?
edit: running this code from the master itself works
As opposed to what was suggested, when connecting to the master using ssh, and running python from the terminal, the very same code (with proper adjustments for the master ip, given it's the same machine) works. No issues no problems.
How does this make sense given the documentation that clearly states otherwise?
You try to run pyspark (which calls spark-submit) form a remote computer outside the spark cluster. This is technically possible but it is not the intended way of deploying applications. In yarn mode, it will make your computer participate in the spark protocol as a client. Thus it would require opening several ports and installing exactly the same spark jars as on spark aws emr.
Form the spark submit doc :
A common deployment strategy is to submit your application from a gateway machine that is physically co-located with your worker machines (e.g. Master node in a standalone EC2 cluster)
A simple deploy strategy is
sync code to master node via rsync, scp or git
cd ~/projects/spark-jobs # on local machine
EMR_MASTER_IP='255.16.17.13'
TARGET_DIR=spark_jobs
rsync -avze "ssh -i ~/dataScienceKey.pem" --rsync-path="mkdir -p ${TARGET_DIR} && rsync" --delete ./ hadoop#${EMR_MASTER_IP}:${TARGET_DIR}
ssh to the master node
ssh -i ~/dataScienceKey.pem hadoop#${EMR_HOST}
run spark-submit on the master node
cd spark_jobs
spark-submit --master yarn --deploy-mode cluster my-job.py
# my-job.py
from pyspark.sql import SparkSession
spark = SparkSession.builder.appName("my-job-py").getOrCreate()
sc = spark.sparkContext
rdd = sc.parallelize([1, 2, 3, 4])
res = rdd.map(lambda x: x**2).collect()
print(res)
There is a way to submit the job directly to spark emr without syncing. Spark EMR runs Apache Livy on port 8998 by default. It is a rest webservice which allows to submit jobs via a rest api. You can pass the same spark-submit parameters with a curl script from your machine. See doc
For interactive development we have also configured local running jupyter notebooks which automatically submit cell runs to livy. This is done via the spark-magic project
According to this Amazon Doc, you can't do that:
Common errors
Standalone mode
Amazon EMR doesn't support standalone mode for Spark. It's not
possible to submit a Spark application to a remote Amazon EMR cluster
with a command like this:
SparkConf conf = new SparkConf().setMaster("spark://master_url:7077”).setAppName("WordCount");
Instead, set up your local machine as explained earlier in this
article. Then, submit the application using the spark-submit command.
You can follow the above linked resource to configure your local machine in order to submit spark jobs to EMR Cluster. Or more simpler, use the ssh key you specified when you create your cluster to connect to the master node and submit spark jobs:
ssh -i ~/path/ssh_key hadoop#$<master_ip_address>
I am new to spark. I want to submit a spark job from local to a remote EMR cluster.
I am following the link here to set up all the prerequisites: https://aws.amazon.com/premiumsupport/knowledge-center/emr-submit-spark-job-remote-cluster/
here is the command as below:
spark-submit --class mymain --deploy-mode client --master yarn myjar.jar
Issue: sparksession creation is not able to be finished with no error. Seems an access issue.
From the aws document, we know that by given the master with yarn, yarn uses the config files I copied from EMR to know where is the master and slaves (yarn-site.xml).
As my EMR cluster is located in a VPC, which need a special ssh config to access, how could I add this info to yarn so it can access to the remote cluster and submit the job?
I think the resolution proposed in aws link is more like - create your local spark setup with all dependencies.
If you don't want to do local spark setup, I would suggest easier way would be, you can use:
1. Livy: for this you emr setup should have livy installed. Check this, this, this and you should be able to infer from this
2. EMR ssh: this requires you to have aws-cli installed locally, cluster id and pem file used while creating emr cluster. Check this
Eg. aws emr ssh --cluster-id j-3SD91U2E1L2QX --key-pair-file ~/.ssh/mykey.pem --command 'your-spark-submit-command' (This prints command output on console though)
I have only a single machine and want to run spark jobs with mesos cluster mode. It might make more sense to run with a cluster of nodes, but I mainly want to test out mesos first to check if it's able to utilize resources more efficiently (run multiple spark jobs at the same time without static partitioning). I have tried a number of ways but without success. Here is what I did:
Build mesos and run both mesos master and slaves (2 slaves in same machines).
sudo ./bin/mesos-master.sh --ip=127.0.0.1 --work_dir=/var/lib/mesos
sudo ./bin/mesos-slave.sh --master=127.0.0.1:5050 --port=5051 --work_dir=/tmp/mesos1
sudo ./bin/mesos-slave.sh --master=127.0.0.1:5050 --port=5052 --work_dir=/tmp/mesos2
Run the spark-mesos-dispatcher
sudo ./sbin/start-mesos-dispatcher.sh --master mesos://localhost:5050
The submit the app with dispatcher as master url.
spark-submit --master mesos://localhost:7077 <other-config> <jar file>
But it doesnt work:
E0925 17:30:30.158846 807608320 socket.hpp:174] Shutdown failed on fd=61: Socket is not connected [57]
E0925 17:30:30.159545 807608320 socket.hpp:174] Shutdown failed on fd=62: Socket is not connected [57]
If I use spark-submit --deploy-mode cluster, then I got another error message:
Exception in thread "main" org.apache.spark.deploy.rest.SubmitRestConnectionException: Unable to connect to server
It work perfectly if I don't use dispatcher but using mesos master url directly: --master mesos://localhost:5050 (client mode). According to the documentation , cluster mode is not supported for Mesos clusters, but they give another instruction for cluster mode here. So it's kind of confusing? My question is:
How I can get it works?
Should I use client mode instead of cluster mode if I submit the app/jar directly from the master node?
If I have a single computer, should I spawn 1 or more mesos slave processes. Basically, I have a number of spark job and dont want to do static partitioning of resources. But when using mesos without static partitioning, it seems to be much slower?
Thanks.
There seem to be two things you're confusing: launching a Spark application in a cluster (as opposed to locally) and launching the driver into the cluster.
From the top of Submitting Applications:
The spark-submit script in Spark’s bin directory is used to launch applications on a cluster. It can use all of Spark’s supported cluster managers through a uniform interface so you don’t have to configure your application specially for each one.
So, Mesos is one of the supported cluster managers and hence you can run Spark apps on a Mesos cluster.
What Mesos as time of writing does not support is launching the driver into the cluster, this is what the command line argument --deploy-mode of ./bin/spark-submitspecifies. Since the default value of --deploy-mode is client you can just omit it, or if you want to explicitly specify it, then use:
./bin/spark-submit --deploy-mode client ...
I use your scenario to try, it could be work.
One thing different , I use ip address to instead of "localhost" and "127.0.0.1"
So just try again and to check http://your_dispatcher:8081 (on browser) if exist.
This is my spark-submit command:
$spark-submit --deploy-mode cluster --master mesos://192.168.11.79:7077 --class "SimpleApp" SimpleAppV2.jar
If success, you can see as below
{
"action" : "CreateSubmissionResponse",
"serverSparkVersion" : "1.5.0",
"submissionId" : "driver-20151006164749-0001",
"success" : true
}
When I got your error log as yours, I reboot the machine and retry your step. It also work.
Try using the 6066 port instead of 7077. The newer versions of Spark prefer the REST api for submitting jobs.
See https://issues.apache.org/jira/browse/SPARK-5388