Running hadoop jar command in edge Node

Question

I am new to hadoop and have below question(s) on running hadoop jar command from edgeNode(http://www.dummies.com/programming/big-data/hadoop/edge-nodes-in-hadoop-clusters/).hadoop jar ${JAR_FILE} {CLASS_NAMEWithPackage} . Have below Question(s)

1. After running above command why the jar is extracted in Djava.io.tmpdir dir in edgeNode ? Every time I run this command I get a directory something like hadoop-unjar7637059002474165348 in temp dir,that has extracted jar.Is this expected? I was thinking hadoop jar submits whole jar to yarn but I could not understand why it is extracted in temp folder ?
2. After extracting the jar in edge Node,does the program expected to remove the extracted jar directory.In this case hadoop-unjar7637059002474165348 ?

Thanks!

Answer 1

You can probably look at here and this question for why your jars are extracted in the edge node (client node) when you run the hadoop jar command. It's to support the 'jar-within-jar' idea while running your jar from the client node. Pushing jars to HDFS, yarn and all those happens after that but, before these happens, your jar has to be executed to begin with, right? In your case, you might have jar-within-jar or you might not, but the concept is supported.

About the auto delete, probably it's not auto deleted.