summaryrefslogtreecommitdiff
path: root/hadoop/streaming
diff options
context:
space:
mode:
Diffstat (limited to 'hadoop/streaming')
-rw-r--r--hadoop/streaming/mapper/input (renamed from hadoop/streaming/mapper_test.input)0
-rw-r--r--hadoop/streaming/mapper/mapper_test.py (renamed from hadoop/streaming/mapper_test.py)0
-rwxr-xr-xhadoop/streaming/mapper/mapper_test.sh (renamed from hadoop/streaming/mapper_test.sh)6
-rw-r--r--hadoop/streaming/mapper/mapper_test1.py (renamed from hadoop/streaming/mapper_test1.py)0
-rwxr-xr-xhadoop/streaming/mapper/mapper_test1.sh (renamed from hadoop/streaming/mapper_test1.sh)8
-rw-r--r--hadoop/streaming/no_reducer/input (renamed from hadoop/streaming/no_reducer.input)0
-rwxr-xr-xhadoop/streaming/no_reducer/no_reducer.rb (renamed from hadoop/streaming/no_reducer.rb)0
-rwxr-xr-xhadoop/streaming/no_reducer/no_reducer.sh (renamed from hadoop/streaming/no_reducer.sh)6
-rw-r--r--hadoop/streaming/partitioner/input (renamed from hadoop/streaming/partitioner_test.input)0
-rwxr-xr-xhadoop/streaming/partitioner/partitioner_test.sh (renamed from hadoop/streaming/partitioner_test.sh)8
-rwxr-xr-xhadoop/streaming/partitioner/partitioner_test1.sh (renamed from hadoop/streaming/partitioner_test1.sh)8
-rw-r--r--hadoop/streaming/secondary_sort/input (renamed from hadoop/streaming/secondary_sort.input)0
-rwxr-xr-xhadoop/streaming/secondary_sort/secondary_sort.sh (renamed from hadoop/streaming/secondary_sort.sh)8
-rw-r--r--hadoop/streaming/test/input (renamed from hadoop/streaming/test.input)0
-rwxr-xr-xhadoop/streaming/test/test.sh (renamed from hadoop/streaming/test.sh)6
15 files changed, 25 insertions, 25 deletions
diff --git a/hadoop/streaming/mapper_test.input b/hadoop/streaming/mapper/input
index 338fd87..338fd87 100644
--- a/hadoop/streaming/mapper_test.input
+++ b/hadoop/streaming/mapper/input
diff --git a/hadoop/streaming/mapper_test.py b/hadoop/streaming/mapper/mapper_test.py
index d358bda..d358bda 100644
--- a/hadoop/streaming/mapper_test.py
+++ b/hadoop/streaming/mapper/mapper_test.py
diff --git a/hadoop/streaming/mapper_test.sh b/hadoop/streaming/mapper/mapper_test.sh
index 4f6e013..f0c5da3 100755
--- a/hadoop/streaming/mapper_test.sh
+++ b/hadoop/streaming/mapper/mapper_test.sh
@@ -1,6 +1,6 @@
#!/bin/sh
-hadoop dfs -put mapper_test.input mapper_test.input
+hadoop dfs -put input input
HADOOP_HOME=/usr/lib/hadoop
HADOOP_VERSION=0.20.2-cdh3u1
@@ -10,7 +10,7 @@ HSTREAMING="$HADOOP_HOME/bin/hadoop jar $HADOOP_HOME/$JAR"
OUT=mapper_test_out
$HSTREAMING \
- -input mapper_test.input \
+ -input input \
-output $OUT \
-mapper "python mapper.py" \
-file mapper_test.py \
@@ -19,5 +19,5 @@ $HSTREAMING \
hadoop dfs -get $OUT .
hadoop dfs -rmr $OUT
-hadoop dfs -rm mapper_test.input
+hadoop dfs -rm input
diff --git a/hadoop/streaming/mapper_test1.py b/hadoop/streaming/mapper/mapper_test1.py
index 79c8aa6..79c8aa6 100644
--- a/hadoop/streaming/mapper_test1.py
+++ b/hadoop/streaming/mapper/mapper_test1.py
diff --git a/hadoop/streaming/mapper_test1.sh b/hadoop/streaming/mapper/mapper_test1.sh
index 80611dc..475699f 100755
--- a/hadoop/streaming/mapper_test1.sh
+++ b/hadoop/streaming/mapper/mapper_test1.sh
@@ -1,17 +1,17 @@
#!/bin/sh
-hadoop dfs -put mapper_test.input mapper_test.input
+hadoop dfs -put input input
HADOOP_HOME=/usr/lib/hadoop
HADOOP_VERSION=0.20.2-cdh3u1
JAR=contrib/streaming/hadoop-streaming-$HADOOP_VERSION.jar
HSTREAMING="$HADOOP_HOME/bin/hadoop jar $HADOOP_HOME/$JAR"
-OUT=mapper_test_out
+OUT=mapper_test1_out
$HSTREAMING \
- -input mapper_test.input \
+ -input input \
-output $OUT \
-mapper "python mapper1.py" \
-file mapper_test1.py \
@@ -20,5 +20,5 @@ $HSTREAMING \
hadoop dfs -get $OUT .
hadoop dfs -rmr $OUT
-hadoop dfs -rm mapper_test.input
+hadoop dfs -rm input
diff --git a/hadoop/streaming/no_reducer.input b/hadoop/streaming/no_reducer/input
index 71ac1b5..71ac1b5 100644
--- a/hadoop/streaming/no_reducer.input
+++ b/hadoop/streaming/no_reducer/input
diff --git a/hadoop/streaming/no_reducer.rb b/hadoop/streaming/no_reducer/no_reducer.rb
index 4410b93..4410b93 100755
--- a/hadoop/streaming/no_reducer.rb
+++ b/hadoop/streaming/no_reducer/no_reducer.rb
diff --git a/hadoop/streaming/no_reducer.sh b/hadoop/streaming/no_reducer/no_reducer.sh
index 7267166..c32bfdd 100755
--- a/hadoop/streaming/no_reducer.sh
+++ b/hadoop/streaming/no_reducer/no_reducer.sh
@@ -1,7 +1,7 @@
#!/bin/sh
-hadoop dfs -put no_reducer.input no_reducer.input
+hadoop dfs -put input input
HADOOP_HOME=/usr/lib/hadoop
HADOOP_VERSION=0.20.2-cdh3u1
@@ -11,7 +11,7 @@ HSTREAMING="$HADOOP_HOME/bin/hadoop jar $HADOOP_HOME/$JAR"
OUT=no_reducer_out
$HSTREAMING \
- -input no_reducer.input \
+ -input input \
-output $OUT \
-mapper "no_reducer.rb" \
-file "no_reducer.rb" \
@@ -19,5 +19,5 @@ $HSTREAMING \
hadoop dfs -get $OUT .
hadoop dfs -rmr $OUT
-hadoop dfs -rm no_reducer.input
+hadoop dfs -rm input
diff --git a/hadoop/streaming/partitioner_test.input b/hadoop/streaming/partitioner/input
index 06c6a28..06c6a28 100644
--- a/hadoop/streaming/partitioner_test.input
+++ b/hadoop/streaming/partitioner/input
diff --git a/hadoop/streaming/partitioner_test.sh b/hadoop/streaming/partitioner/partitioner_test.sh
index dcc7353..bfb2185 100755
--- a/hadoop/streaming/partitioner_test.sh
+++ b/hadoop/streaming/partitioner/partitioner_test.sh
@@ -1,22 +1,22 @@
#!/bin/sh
-hadoop dfs -put partitioner_test.input partitioner_test.input
+hadoop dfs -put input input
HADOOP_HOME=/usr/lib/hadoop
HADOOP_VERSION=0.20.2-cdh3u1
JAR=contrib/streaming/hadoop-streaming-$HADOOP_VERSION.jar
HSTREAMING="$HADOOP_HOME/bin/hadoop jar $HADOOP_HOME/$JAR"
-OUT=partitioner_out
+OUT=partitioner_test_out
$HSTREAMING \
- -input partitioner_test.input \
+ -input input \
-output $OUT \
-mapper /bin/cat \
-jobconf mapred.reduce.tasks=2
hadoop dfs -get $OUT .
hadoop dfs -rmr $OUT
-hadoop dfs -rm partitioner_test.input
+hadoop dfs -rm input
diff --git a/hadoop/streaming/partitioner_test1.sh b/hadoop/streaming/partitioner/partitioner_test1.sh
index 6e6344c..759f823 100755
--- a/hadoop/streaming/partitioner_test1.sh
+++ b/hadoop/streaming/partitioner/partitioner_test1.sh
@@ -1,17 +1,17 @@
#!/bin/sh
-hadoop dfs -put partitioner_test.input partitioner_test.input
+hadoop dfs -put input input
HADOOP_HOME=/usr/lib/hadoop
HADOOP_VERSION=0.20.2-cdh3u1
JAR=contrib/streaming/hadoop-streaming-$HADOOP_VERSION.jar
HSTREAMING="$HADOOP_HOME/bin/hadoop jar $HADOOP_HOME/$JAR"
-OUT=partitioner1_out
+OUT=partitioner_test1_out
$HSTREAMING \
- -input partitioner_test.input \
+ -input input \
-output $OUT \
-mapper /bin/cat \
-jobconf mapred.reduce.tasks=2 \
@@ -23,5 +23,5 @@ $HSTREAMING \
hadoop dfs -get $OUT .
hadoop dfs -rmr $OUT
-hadoop dfs -rm partitioner_test.input
+hadoop dfs -rm input
diff --git a/hadoop/streaming/secondary_sort.input b/hadoop/streaming/secondary_sort/input
index 5aa7ec9..5aa7ec9 100644
--- a/hadoop/streaming/secondary_sort.input
+++ b/hadoop/streaming/secondary_sort/input
diff --git a/hadoop/streaming/secondary_sort.sh b/hadoop/streaming/secondary_sort/secondary_sort.sh
index 7fa0c6d..c45f381 100755
--- a/hadoop/streaming/secondary_sort.sh
+++ b/hadoop/streaming/secondary_sort/secondary_sort.sh
@@ -1,17 +1,17 @@
#!/bin/sh
-hadoop dfs -put secondary_sort.input secondary_sort.input
+hadoop dfs -put input input
HADOOP_HOME=/usr/lib/hadoop
HADOOP_VERSION=0.20.2-cdh3u1
JAR=contrib/streaming/hadoop-streaming-$HADOOP_VERSION.jar
HSTREAMING="$HADOOP_HOME/bin/hadoop jar $HADOOP_HOME/$JAR"
-OUT=comp_out
+OUT=secondary_sort_out
$HSTREAMING \
- -input secondary_sort.input \
+ -input input \
-output $OUT \
-partitioner org.apache.hadoop.mapred.lib.KeyFieldBasedPartitioner \
-jobconf map.output.key.field.separator="-*-" \
@@ -26,5 +26,5 @@ $HSTREAMING \
hadoop dfs -get $OUT .
hadoop dfs -rmr $OUT
-hadoop dfs -rm secondary_sort.input
+hadoop dfs -rm input
diff --git a/hadoop/streaming/test.input b/hadoop/streaming/test/input
index 338fd87..338fd87 100644
--- a/hadoop/streaming/test.input
+++ b/hadoop/streaming/test/input
diff --git a/hadoop/streaming/test.sh b/hadoop/streaming/test/test.sh
index 3cb47d5..2f09bac 100755
--- a/hadoop/streaming/test.sh
+++ b/hadoop/streaming/test/test.sh
@@ -1,7 +1,7 @@
#!/bin/sh
-hadoop dfs -put test.input test.input
+hadoop dfs -put input input
HADOOP_HOME=/usr/lib/hadoop
HADOOP_VERSION=0.20.2-cdh3u1
@@ -11,7 +11,7 @@ HSTREAMING="$HADOOP_HOME/bin/hadoop jar $HADOOP_HOME/$JAR"
OUT=test_out
$HSTREAMING \
- -input test.input \
+ -input input \
-output $OUT \
-mapper /bin/cat \
-jobconf "mapred.reduce.tasks=3" \
@@ -19,5 +19,5 @@ $HSTREAMING \
hadoop dfs -get $OUT .
hadoop dfs -rmr $OUT
-hadoop dfs -rm test.input
+hadoop dfs -rm input
generated by cgit v1.2.3 (git 2.39.1) at 2024-11-22 08:43:24 +0000