Best to use a secured Hadoop cluster [0], and/or setup appropriate firewall rules that block traffic from other than your trusted IPs.
[0] - https://ccp.cloudera.com/display/CDHDOC/CDH3+Security+Guide On Mon, Jan 16, 2012 at 4:33 AM, Something Something <mailinglist...@gmail.com> wrote: > Good point. Those ports may not be open. So next question - is it safe to > open these ports? How do we securely open these ports to avoid malicious > attacks under EC2? > > (Sorry, I know some of these questions are dumb - but we are a startup and > don't have a big sysadmin group - I guess that's why we are using EC2 -:) > > > On Sun, Jan 15, 2012 at 12:03 PM, Ronald Petty <ronald.pe...@gmail.com> > wrote: >> >> Something Something, >> >> Have you confirmed you can connect to the port from your remote machine? >> >> telnet ec2-xxxxxx.... 9000 >> >> Kindest regards. >> >> Ron >> >> On Sun, Jan 15, 2012 at 12:16 AM, Something Something >> <mailinglist...@gmail.com> wrote: >>> >>> Hello, >>> >>> Our Hadoop cluster is setup on EC2, but our client machine which will >>> trigger the M/R job is in our data center. I am trying to start a M/R job >>> from our client machine, but getting this: >>> >>> 00:01:16.885 [pool-6-thread-1] INFO org.apache.hadoop.ipc.Client - >>> Retrying connect to server: >>> ec2-xx-xx-xxx-xxx.compute-1.amazonaws.com/xx.xx.xxx.xx:9000. Already tried 0 >>> time(s). >>> 00:01:16.885 [pool-6-thread-1] INFO org.apache.hadoop.ipc.Client - >>> Retrying connect to server: >>> ec2-xx-xx-xxx-xxx.compute-1.amazonaws.com/xx.xx.xxx.xx:9000. Already tried 1 >>> time(s). >>> 00:01:16.885 [pool-6-thread-1] INFO org.apache.hadoop.ipc.Client - >>> Retrying connect to server: >>> ec2-xx-xx-xxx-xxx.compute-1.amazonaws.com/xx.xx.xxx.xx:9000. Already tried 2 >>> time(s). >>> & so on... >>> >>> >>> I tried putting our 'pem' file in the classpath, but that didn't help. I >>> am assuming it should be possible to trigger a job on EC2 cluster from a >>> machine that's not in EC2, correct? >>> >>> Please note that I can trigger a M/R job from the client machine to a >>> Hadoop running on the same machine - localhost:9000 (hdfs) as well as >>> localhost:9001 (jobtracker). >>> >>> Please help. Thanks. >>> >>> >>> >>> >> > -- Harsh J Customer Ops. Engineer, Cloudera
