def code():
run_jar(
"%s/flink-jobs/target" % generators.ALS.repo.get_absolute_path(),
"flink-jobs-*.jar",
args=["master", 15, 1, 10, 100, "rand", "%s/als-temp/" % get_hdfs_address(), als_in, als_out],
clazz="com.github.projectflink.als.ALSJoinBlocking",
)
def code():
run_jar("%s/target/" % self.repo.get_absolute_path(),
"flink-dataflow-*-SNAPSHOT.jar",
args = ["--", # Flink 0.8 way of specifying options to user programs
"--input=%s" % wordcount_in,
"--output=%s" % self.wordcount_out],
clazz = "com.dataartisans.flink.dataflow.examples.DataflowWordCount")
def code():
run_jar(
"%s/target/" % self.repo.get_absolute_path(),
"flink-dataflow-*-SNAPSHOT.jar",
args=[wordcount_in, self.wordcount_out],
clazz=
"com.dataartisans.flink.dataflow.examples.StreamingPipeline")
def code():
run_jar(
"%s/examples/streaming/" % get_flink_dist_path(),
"WindowWordCount.jar",
args=[self.wordcount_in, self.wordcount_out, 10000],
clazz=
"org.apache.flink.streaming.examples.windowing.WindowWordCount"
)
def code():
run_jar("%s/examples/batch/" % get_flink_dist_path(),
"WordCount.jar",
args=[
"--input", self.wordcount_in, "--output",
self.wordcount_out
],
clazz="org.apache.flink.examples.java.wordcount.WordCount")
def code():
run_jar(path = "experiments/wordcount_files/",
jar_name = "flink-java-examples-0.8-incubating-SNAPSHOT-WordCount.jar",
args = [
"hdfs://%s:50040/generated-wc.txt" % env.master,
"hdfs://%s:50040/tmp/wc-out/" % env.master
],
upload=True
)
def code():
run_jar("%s/flink-jobs/target" % self.repo.get_absolute_path(),
"flink-jobs-*.jar",
args=[
self.num_rows, self.num_cols,
self.mean_entry, self.variance_entry,
self.mean_num_row_entries, self.variance_num_row_entries,
self.out_path
],
clazz="com.github.projectflink.als.ALSDataGeneration")
def code():
run_jar("%s/flink-jobs/target" % self.repo.get_absolute_path(),
"flink-jobs-*.jar",
args=[
"-s", self.scale_factor,
"-p", self.parallelism,
"-o", self.out,
],
clazz="com.github.projectflink.avro.GenerateLineitems"
)
def code():
run_jar(
path="experiments/wordcount_files/",
jar_name=
"flink-java-examples-0.8-incubating-SNAPSHOT-WordCount.jar",
args=[
"hdfs://%s:50040/generated-wc.txt" % env.master,
"hdfs://%s:50040/tmp/wc-out/" % env.master
],
upload=True)
def code():
run_jar("%s/flink-jobs/target" % self.repo.get_absolute_path(),
"flink-jobs-*.jar",
args=[
"-s", self.scale_factor,
"-p", self.parallelism,
"-o", self.out,
],
clazz="com.github.projectflink.avro.GenerateLineitems"
)
def code():
run_jar("%s/flink-jobs/target" % self.repo.get_absolute_path(),
"flink-jobs-*.jar",
args=[
self.num_rows, self.num_cols,
self.mean_entry, self.variance_entry,
self.mean_num_row_entries, self.variance_num_row_entries,
self.out_path
],
clazz="com.github.projectflink.als.ALSDataGeneration")
def code():
run_jar("%s/flink-jobs/target" % generators.ALS.repo.get_absolute_path(),
"flink-jobs-*.jar",
args = [
"master",
15, 1, 10, 100, "rand",
"%s/als-temp/" % get_hdfs_address(),
als_in,
als_out
],
clazz = "com.github.projectflink.als.ALSJoinBlocking")
def code():
run_jar(
"%s/target/" % self.repo.get_absolute_path(),
"flink-dataflow-*-SNAPSHOT.jar",
args=[
"--", # Flink 0.8 way of specifying options to user programs
"--input=%s" % wordcount_in,
"--output=%s" % self.wordcount_out
],
clazz=
"com.dataartisans.flink.dataflow.examples.DataflowWordCount")
def code():
run_jar("~/flink-perf/flink-jobs/target",
"flink-jobs-*.jar",
args=[grep_in, grep_out, "these", "are", "test", "words"],
clazz="com.github.projectflink.grep.GrepJob")
def code():
run_jar("%s/flink-jobs/target" % self.repo.get_absolute_path(),
"flink-jobs-*.jar",
args=[self.dop, self.out_path, self.size_gb],
clazz="com.github.projectflink.generators.Text")
def code():
run_jar("%s/examples/streaming/" % get_flink_dist_path(),
"WordCount.jar",
args = ["--input", self.wordcount_in, "--output", self.wordcount_out],
clazz = "org.apache.flink.streaming.examples.wordcount.WordCount")
def code():
run_jar("%s/target/" % self.repo.get_absolute_path(),
"flink-dataflow-*-SNAPSHOT.jar",
args = [wordcount_in, self.wordcount_out],
clazz = "com.dataartisans.flink.dataflow.GoogleStreamingPipeline.examples")
def code():
run_jar("%s/target/" % self.repo.get_absolute_path(),
"flink-dataflow-*-SNAPSHOT.jar",
args = [wordcount_in, self.wordcount_out],
clazz = self.implicit_clazz if self.implicit_combine else self.explicit_clazz)
def code():
run_jar("%s/flink-staging/flink-streaming/flink-streaming-examples/target/" % get_flink_path(),
"flink-streaming-*-WordCount.jar",
args = [self.wordcount_in, self.wordcount_out],
clazz = "org.apache.flink.streaming.examples.wordcount.WordCount")
def code():
run_jar("%s/examples/streaming/" % get_flink_dist_path(),
"WindowWordCount.jar",
args = [self.wordcount_in, self.wordcount_out, 10000],
clazz = "org.apache.flink.streaming.examples.windowing.WindowWordCount")
def code():
run_jar("%s/examples/" % get_flink_dist_path(),
"flink-java-*WordCount.jar",
args = [wordcount_out, wordcount_in],
clazz = "org.apache.flink.examples.java.wordcount.WordCount")
def code():
run_jar("%s/flink-jobs/target" % self.repo.get_absolute_path(),
"flink-jobs-*.jar",
args=[self.dop, self.out_path, self.size_gb],
clazz="com.github.projectflink.generators.Text")
def code():
run_jar("%s/target/" % self.repo.get_absolute_path(),
"flink-dataflow-*-SNAPSHOT.jar",
args=[wordcount_in, self.wordcount_out],
clazz=self.implicit_clazz
if self.implicit_combine else self.explicit_clazz)
def code():
run_jar("~/flink-perf/flink-jobs/target",
"flink-jobs-*.jar",
args = [grep_in, grep_out, "these", "are", "test", "words"],
clazz = "com.github.projectflink.grep.GrepJob")
def code():
run_jar("%s/flink-jobs/target" %
generators.Avro.repo.get_absolute_path(),
"flink-jobs-*.jar",
args=[self.out_path, self.in_path],
clazz="com.github.projectflink.avro.CompareJob")
