We have a mesosphere cluster running on EC2 with hdfs service installed. I was able to issue "hadoop jar" command to start a mapreduce job. And it looks like the installation comes with full CDH5 distributions.
This makes us thinking if it is possible to run hadoop mapreduce job with this setting. If this is doable, what is the right practice to setup?
Glad to hear it's working for you @JingzhouLuo! MR-1 should work. If you want to run MR-2 (Yarn), I suggest checking out Myriad (www.github.com/mesos/myriad) which runs on top of HDFS.
Hi,
We have a mesosphere cluster running on EC2 with hdfs service installed. I was able to issue "hadoop jar" command to start a mapreduce job. And it looks like the installation comes with full CDH5 distributions.
This makes us thinking if it is possible to run hadoop mapreduce job with this setting. If this is doable, what is the right practice to setup?
Thanks, Jingzhou