Configuring Network Names
To ensure that the members of the cluster can communicate with each other, do the following on every system:
- Set the hostname of each system to a unique name (not localhost). For
$ sudo hostname myhost-1Note: This is a temporary measure only. The hostname set by hostname does not survive across reboots
- Make sure the /etc/hosts file on each system contains the IP addresses
and fully-qualified domain names (FQDN) of all the members of the cluster.
The canonical name of each host in /etc/hosts must be the FQDN (for example myhost-1.mynet.myco.com), not the unqualified hostname (for example myhost-1). The canonical name is the first entry after the IP address.
If you are using DNS, storing this information in /etc/hosts is not required, but it is good practice.
- Make sure the /etc/sysconfig/network file on each system contains the hostname you have just set (or verified) for that system, for example myhost-1.
- Check that this system is consistently identified to the network:
- Run uname -a and check that the hostname matches the output of the hostname command.
- Run /sbin/ifconfig and note the value of inet addr
in the eth0 entry, for example:
$ /sbin/ifconfig eth0 Link encap:Ethernet HWaddr 00:0C:29:A4:E8:97 inet addr:172.29.82.176 Bcast:172.29.87.255 Mask:255.255.248.0 ...
- Run host -v -t A `hostname` and make sure that hostname matches the
output of the hostname command, and has the same IP address as reported
by ifconfig for eth0; for example:
$ host -v -t A `hostname` Trying "myhost.mynet.myco.com" ... ;; ANSWER SECTION: myhost.mynet.myco.com. 60 IN A 172.29.82.176
- For MRv1: make sure conf/core-site.xml and conf/mapred-site.xml, respectively, have the hostnames – not the IP addresses – of the NameNode and the JobTracker. These can be FQDNs (for example myhost-1.mynet.myco.com), or unqualified hostnames (for example myhost-1). See Customizing Configuration Files and Deploying MapReduce v1 (MRv1) on a Cluster.
- For YARN: make sure conf/core-site.xml and conf/yarn-site.xml, respectively, have the hostnames – not the IP addresses – of the NameNode, the ResourceManager, and the ResourceManager Scheduler. See Customizing Configuration Files and Deploying MapReduce v2 (YARN) on a Cluster.
- Make sure that components that depend on a client-server relationship – Oozie, HBase, ZooKeeper – are configured according to the instructions on their installation pages:
|<< Previous: Deploying CDH4 on a Cluster||Next: Deploying HDFS on a Cluster >>|