Hello everyone,
I am working on setting up a cluster using Ambari, VMWare (Big Data Extensions), and EMC Isilon for HDFS storage. I have starter kit documentation from EMC that states that Ambari requires a fairly large root partition on the OS, in the documentation they are saying to give it 250GB. This has caused some concern for our Linux administrators and I haven’t found a definitive best practice regarding the / partition size. For slave nodes the documentation is saying 20GB (this is also the default size for the VM when done with BDE) is recommended, but I need to know the best practice for the Ambari node. If the overall size of the cluster has an impact, this will be 1 Ambari node (VM) 3 MapReduce nodes (VM’s) & 3 Name/Data nodes (Isilon) for the size for our POC.
Does Ambari really require such a large root partition (250GB)? If so, can I direct the data that would normally have been stored there in a separate partition and change that in a configuration?
Thanks in advance for any recommendations,
Brett