I have successfully deployed a cluster using Cloudbreak. I used the supplied hadoop-spark blueprint. However I can't find any way to issue hadoop commands (i.e. "hadoop fsck / ")
I've run "which hadoop" on each VM and none of them returns any hadoop location.
What am I missing here? How do I successfully issue hadoop commands on my cluster?
# list docker containers
# connect to container that has ambari-agent
docker exec -it <ambari agent docker id> bash or sudo docker exec -it $(sudo docker ps |grep ambari- | cut -f 1 -d " ") bash
Then run your hadoop commands
@Ancil McBarnett we have some special commands for the easier usage like ambari-enter which enters into the ambari container. Read more here: http://sequenceiq.com/cloudbreak-docs/release-1.1.0/operations/#cloudbreak-application
Which machine should these commands be run on? I am running the supplied hadoop-spark blueprint using 4 slaves. The blueprint set up the following machines for me:
This is my result from running on cbgateway-0, so it seems this is not the right machine. Which leads to the question , which one is correct?
you'll need to run this on the machine hdfs client is installed on. Look at your Ambari install and find a machine that has clients, worst case, run the docker command on every node until you find the one. @Thom Rogers
@Thom Rogers the cbgateway-0 contains the ambari-server and all other machine contains the ambari agents. The machine which is needs for you depends on the blueprint so If you want to use hdfs then you need a hdfs-client service. If the configuration of the hostgroupslave1 contains an hdfs-client in your blueprint then every machine which name is hostgroupslave1* contains an installed hdfs-client.