Hadoop provides a feature that lets administrators specify mapping rules to map a kerberos principal to a local UNIX user name.This required with Kerberized Hadoop clusters to turn full UPN’s into the shortnames required by the HDFS services. In Ambari these rules look similar to this and are added to the core-site.xml: Kerberized Ambari […]Read more "Hadoop’s hadoop.security.auth_to_local rules"
if you are seeing issues with kerberos based hdfs client access against an Isilon cluster, increasing the kerberos logging level on the client can show you a lot more information(a lot) Change the logging level with the following: export HADOOP_ROOT_LOGGER=”TRACE,console” export HADOOP_OPTS=”-Dsun.security.krb5.debug=true” quick example of a successful kerberized hadoop call from a kerberized […]Read more "Increase the kerberos logging level on a client to see what is actually going on"
I’m starting to see an increase in the amount of Kerberized access to Isilon clusters, the following is an excellent start to understanding the process of Kerberizing an Isilon cluster for NFS access. In later posts I’ll discuss Kerberized Hadoop access to Isilon. But, this doc is a great first step, it’s hosted on […]Read more "Isilon and Kerberos"
Wanted to clear up the implementation of racks with Isilon hdfs and providing rack awareness, using Isilon racks we can provide node location awareness to clients and within-switch datanode connections collocation. This gives you the capability to provide switch aware or top of rack functionality that emulates hdfs local storage implementations, where compute nodes attempt […]Read more "Isilon HDFS and Rack Awareness for DataNode Connections"
You will very likely see this blog focused on implementing big data solutions on EMC Isilon mostly, but occasionally some ScaleIO content will get posted. These are the two products that I spend most of my time working with and implementing currently. But, let’s see what happens…Read more "what this blog will be about"
There appears to be some confusion here; the older Isilon Hadoop docs recommended separate subnet pools for the NameNode (NN) and the DataNodes (DN), this changed recently and now the suggested recommendation is a single subnet pool for NameNode (NN) and DataNodes (DN), the pool should be dynamic to provide clean failover of HDFS services. […]Read more "How many SmartConnect Zones on Isilon for HDFS"
I’m starting up a blog to publish anything useful I write on technology implementations and storage architectures. Hopefully someone finds this content useful sometime. Enjoy!Read more "Getting Started"