|
Last change
on this file since 31258 was 31258, checked in by davidb, 7 years ago |
|
POS Label count, similar to Whitelist word count
|
-
Property svn:executable
set to
*
|
|
File size:
621 bytes
|
| Line | |
|---|
| 1 | #!/bin/bash
|
|---|
| 2 |
|
|---|
| 3 | json_filelist=${1:-full-listing-step100000.txt}
|
|---|
| 4 | shift
|
|---|
| 5 |
|
|---|
| 6 | input_dir="hdfs://$SPARK_MASTER_HOST:9000/tmp/dbbridge/full-ef-json-files"
|
|---|
| 7 | #input_dir="hdfs://master:9000/user/htrc/full-ef-json-files"
|
|---|
| 8 | #input_dir="hdfs://10.10.0.52:9000/user/htrc/full-ef-json-files"
|
|---|
| 9 |
|
|---|
| 10 | #output_dir=hdfs://master:9000/user/htrc/full-solr-json-files
|
|---|
| 11 | solr_url="http://gc0:8983/solr/htrc-full-ef/update"
|
|---|
| 12 |
|
|---|
| 13 | #
|
|---|
| 14 |
|
|---|
| 15 | master_opt="--executor-memory 10g --conf spark.executor.extraJavaOptions=-XX:+HeapDumpOnOutOfMemoryError --master spark://$SPARK_MASTER_HOST:7077"
|
|---|
| 16 |
|
|---|
| 17 | classmain="org.hathitrust.extractedfeatures.ProcessForPOSCount"
|
|---|
| 18 |
|
|---|
| 19 | . scripts/_RUN.sh
|
|---|
| 20 |
|
|---|
Note:
See
TracBrowser
for help on using the repository browser.
