diff options
Diffstat (limited to 'nixpkgs/nixos/tests/spark')
-rw-r--r-- | nixpkgs/nixos/tests/spark/default.nix | 48 | ||||
-rw-r--r-- | nixpkgs/nixos/tests/spark/spark_sample.py | 40 |
2 files changed, 88 insertions, 0 deletions
diff --git a/nixpkgs/nixos/tests/spark/default.nix b/nixpkgs/nixos/tests/spark/default.nix new file mode 100644 index 000000000000..034e9711bed5 --- /dev/null +++ b/nixpkgs/nixos/tests/spark/default.nix @@ -0,0 +1,48 @@ +{ pkgs, ... }: + +let + inherit (pkgs) lib; + tests = { + default = testsForPackage { sparkPackage = pkgs.spark; }; + }; + + testsForPackage = args: lib.recurseIntoAttrs { + sparkCluster = testSparkCluster args; + passthru.override = args': testsForPackage (args // args'); + }; + testSparkCluster = { sparkPackage, ... }: pkgs.testers.nixosTest ({ + name = "spark"; + + nodes = { + worker = { nodes, pkgs, ... }: { + services.spark = { + package = sparkPackage; + worker = { + enable = true; + master = "master:7077"; + }; + }; + virtualisation.memorySize = 2048; + }; + master = { config, pkgs, ... }: { + services.spark = { + package = sparkPackage; + master = { + enable = true; + bind = "0.0.0.0"; + }; + }; + networking.firewall.allowedTCPPorts = [ 22 7077 8080 ]; + }; + }; + + testScript = '' + master.wait_for_unit("spark-master.service") + worker.wait_for_unit("spark-worker.service") + worker.copy_from_host( "${./spark_sample.py}", "/spark_sample.py" ) + assert "<title>Spark Master at spark://" in worker.succeed("curl -sSfkL http://master:8080/") + worker.succeed("spark-submit --version | systemd-cat") + worker.succeed("spark-submit --master spark://master:7077 --executor-memory 512m --executor-cores 1 /spark_sample.py") + ''; + }); +in tests diff --git a/nixpkgs/nixos/tests/spark/spark_sample.py b/nixpkgs/nixos/tests/spark/spark_sample.py new file mode 100644 index 000000000000..c4939451eae0 --- /dev/null +++ b/nixpkgs/nixos/tests/spark/spark_sample.py @@ -0,0 +1,40 @@ +from pyspark.sql import Row, SparkSession +from pyspark.sql import functions as F +from pyspark.sql.functions import udf +from pyspark.sql.types import * +from pyspark.sql.functions import explode + +def explode_col(weight): + return int(weight//10) * [10.0] + ([] if weight%10==0 else [weight%10]) + +spark = SparkSession.builder.getOrCreate() + +dataSchema = [ + StructField("feature_1", FloatType()), + StructField("feature_2", FloatType()), + StructField("bias_weight", FloatType()) +] + +data = [ + Row(0.1, 0.2, 10.32), + Row(0.32, 1.43, 12.8), + Row(1.28, 1.12, 0.23) +] + +df = spark.createDataFrame(spark.sparkContext.parallelize(data), StructType(dataSchema)) + +normalizing_constant = 100 +sum_bias_weight = df.select(F.sum('bias_weight')).collect()[0][0] +normalizing_factor = normalizing_constant / sum_bias_weight +df = df.withColumn('normalized_bias_weight', df.bias_weight * normalizing_factor) +df = df.drop('bias_weight') +df = df.withColumnRenamed('normalized_bias_weight', 'bias_weight') + +my_udf = udf(lambda x: explode_col(x), ArrayType(FloatType())) +df1 = df.withColumn('explode_val', my_udf(df.bias_weight)) +df1 = df1.withColumn("explode_val_1", explode(df1.explode_val)).drop("explode_val") +df1 = df1.drop('bias_weight').withColumnRenamed('explode_val_1', 'bias_weight') + +df1.show() + +assert(df1.count() == 12) |