This is an automated email from the ASF dual-hosted git repository. snagel pushed a commit to branch master in repository https://gitbox.apache.org/repos/asf/nutch.git
commit e41e55afcbfcccd6f2fa8ddae3c2137a9fdd122b Merge: 50c6c23 4475878 Author: Sebastian Nagel <sna...@apache.org> AuthorDate: Thu Apr 26 13:13:53 2018 +0200 Merge pull request #323 from sebastian-nagel/NUTCH-2570-dedup-job-merge-crawldb NUTCH-2570 Deduplication job fails to install deduplicated CrawlDb .../org/apache/nutch/crawl/DeduplicationJob.java | 26 +++++++++++++--------- 1 file changed, 15 insertions(+), 11 deletions(-) diff --cc src/java/org/apache/nutch/crawl/DeduplicationJob.java index eaeb835,12ebd3c..8887b4f --- a/src/java/org/apache/nutch/crawl/DeduplicationJob.java +++ b/src/java/org/apache/nutch/crawl/DeduplicationJob.java @@@ -293,12 -292,11 +292,12 @@@ public class DeduplicationJob extends N Job job = NutchJob.getInstance(getConf()); Configuration conf = job.getConfiguration(); - job.setJobName("Deduplication on " + crawldb); + job.setJobName("Deduplication on " + crawlDb); conf.set(DEDUPLICATION_GROUP_MODE, group); conf.set(DEDUPLICATION_COMPARE_ORDER, compareOrder); + job.setJarByClass(DeduplicationJob.class); - FileInputFormat.addInputPath(job, new Path(crawldb, CrawlDb.CURRENT_NAME)); + FileInputFormat.addInputPath(job, new Path(crawlDb, CrawlDb.CURRENT_NAME)); job.setInputFormatClass(SequenceFileInputFormat.class); FileOutputFormat.setOutputPath(job, tempDir); -- To stop receiving notification emails like this one, please contact sna...@apache.org.