Charmed Spark K8s

  • Canonical | bundle
Channel Revision Published
latest/edge 4 06 Aug 2024
3.4/edge 4 06 Aug 2024
juju deploy spark-k8s-bundle --channel 3.4/edge
Show information

Platform:

Interacting with Spark using Interactive Shell

Spark comes with an interactive shell that provides a simple way to learn the API. It is available in either Scala or Python. In this section, we’re going to play around a bit with Spark’s shell to interact directly with the Spark cluster.

PySpark Shell

Spark comes with a built-in Python shell where we can execute commands interactively against the Spark cluster using the Python programming language.

PySpark shell can be launched with the spark-client snap like this:

spark-client.pyspark \
  --username spark --namespace spark

Once the shell is open and ready, you should see a prompt simlar to the following:

Welcome to
      ____              __
     / __/__  ___ _____/ /__
    _\ \/ _ \/ _ `/ __/  '_/
   /__ / .__/\_,_/_/ /_/\_\   version 3.4.1
      /_/

Using Python version 3.10.12 (main, Jun 11 2023 05:26:28)
Spark context Web UI available at http://172.31.29.134:4040
Spark context available as 'sc' (master = k8s://https://172.31.29.134:16443, app id = spark-b364a0a9a2cb41f3a713c62b900fbd82).
SparkSession available as 'spark'.
>>> 

When you open the PySpark shell, Spark spawns a couple of executor pods in the background to process commands. You can see them by fetching the list of pods in the spark namespace in a separate shell.

kubectl get pods -n spark

You should see output lines similar to the following:

pysparkshell-xxxxxxxxxxxxxxxx-exec-1              1/1     Running            0          xs
pysparkshell-xxxxxxxxxxxxxxxx-exec-2              1/1     Running            0          xs

As you can see, PySpark spawned two executor pods within the spark namespace. This is the namespace that we provided as a value to the --namespace argument when launching pyspark. It’s in these executor pods that data is cached and the computation will be executed, therefore creating a computational architecture that can horizontally scale to large datasets (“big data”). On the other hand, the PySpark shell started by the spark-client snap will act as a driver, controlling and orchestrating the operations of the executors. More information about the Spark architecture can be found here.

One good thing about the shell is that the Spark context and session is already pre-loaded onto the shell and can be easily accessed with variables sc and spark respectively. You can even see this printed in the logs above, where upon initialization, the PySpark shell says Spark context available as 'sc' and SparkSession available as 'spark'. This shell is just like a regular Python shell, with Spark context loaded on top of it.

To start, you can print ‘hello, world!’ just like you’d do in a Python shell.

>>> print('hello, world!')

hello, world!

Let’s try a simple example of counting the number of vowel characters in a string. The following is the string that we are going to use:

lines = """Canonical's Charmed Data Platform solution for Apache Spark runs Spark jobs on your Kubernetes cluster.
You can get started right away with MicroK8s - the mightiest tiny Kubernetes distro around! 
The spark-client snap simplifies the setup process to get you running Spark jobs against your Kubernetes cluster. 
Spark on Kubernetes is a complex environment with many moving parts.
Sometimes, small mistakes can take a lot of time to debug and figure out.
"""

The following is a function that returns the number of vowel characters in the string:

def count_vowels(text: str) -> int:
  count = 0
  for char in text:
    if char.lower() in "aeiou":
      count += 1
  return count

To test this function, the string lines can now be passed into it and the number of vowels is printed to the console as follows:

>>> count_vowels(lines)
128

Since Spark is a distributed processing framework, we can split up this task and parallellize it over multiple executor pods. This parallelization can be done as simply as:

>>> from operator import add
>>> spark.sparkContext.parallelize(lines.splitlines(), 2).map(count_vowels).reduce(add)
128

Here, we split the data into the two executors (passed as an argument to parallelize function), generating a distributed data structure, e.g. RDD[str], where each line is stored in one of the (possibly many) executors. The number of vowels in each line is then computed, line by line, with the map function, and then the numbers are aggregated and added up to calculate the total number of occurrences of vowel characters in the entire dataset. This kind of parallelization of tasks is particularly useful in processing very large data sets which helps to reduce the processing time significantly, and it is generally referred to as the MapReduce pattern.

To exit from PySpark shell, you can simply run exit() or press Ctrl + Z key combination.

Scala Shell

Spark comes with a built-in interactive Scala shell as well. Enter the following command to enter an interactive Scala shell:

spark-client.spark-shell --username spark --namespace spark --num-executors 4

Notice how we can specify the number of executors when initializing the Spark Shell (which by default would be 2). Once the shell is open and ready, you should see a prompt simlar to the following:

Welcome to
      ____              __
     / __/__  ___ _____/ /__
    _\ \/ _ \/ _ `/ __/  '_/
   /___/ .__/\_,_/_/ /_/\_\   version 3.4.1
      /_/
         
Using Scala version 2.12.17 (OpenJDK 64-Bit Server VM, Java 11.0.20.1)
Type in expressions to have them evaluated.
Type :help for more information.

scala> 

Just as in the PySpark shell, the spark context and spark session are readily available in the shell as sc and spark respectively. Moreover, new executor pods are created in spark namespace in order to execute the commands, just like with the PySpark shell.

The example of counting the vowel characters can be equivalently run in Scala with the following lines:

val lines = """Canonical's Charmed Data Platform solution for Apache Spark runs Spark jobs on your Kubernetes cluster.
You can get started right away with MicroK8s - the mightiest tiny Kubernetes distro around! 
The spark-client snap simplifies the setup to run Spark jobs against your Kubernetes cluster. 
Spark on Kubernetes is a complex environment with many moving parts.
Sometimes, small mistakes can take a lot of time to debug and figure out.
"""

def countVowels(text: String): Int = {
  text.toLowerCase.count("aeiou".contains(_))
}

sc.parallelize(lines.split("\n"), 2).map(countVowels).reduce(_ + _)

To exit from the Scala shell, simply press Ctrl + C key combination.

Interactive shells are a great way to try out experiments and to learn the basics of the Spark. For a more advanced use case, jobs can be submitted to the Spark cluster as scripts using spark-submit. That’s what we will do in the next section.


Help improve this document in the forum (guidelines). Last updated 8 months ago.