08 Mar / 2012
Getting DNS right for Hadoop / HBase clusters
Hadoop and HBase (especially HBase) are very picky about DNS
entries. When setting up a Hadoop cluster one doesn’t
always have access to a DNS server. So here is ‘poor
developers’ guide to getting DNS correct.
Following these simple steps, can avoid a few thorny issues down the
line.
- set Hostname
- verify hostname –> IP address resolution is working (DNS
resolution) - verify IP address –> hostname resolution is working (reverse
DNS) - DNS verification tool
1) Hostname
I like to set these to FULLY QUALIFIED NAMES.
so ‘ hadoop1.lab.mycompany.com‘
is good
just ‘hadoop1’ is not.
on CENTOS
set this in ‘/etc/sysconfig/network‘
HOSTNAME=hadoop1.lab.mycompany.com
on UBUNTU:
set this on ‘/etc/hostname‘
hadoop1.lab.mycompany.com
just reboot the host for hostname settings to take effect (to be safe)
Do this at every node.
2) DNS entries when you don’t have DNS server
So don’t want to mess around (or can’t) with DNS server? No
worries. We can use ‘/etc/hosts’ file to make our own
tiny DNS for our hadoop cluster
file : /etc/hosts
add the following *AFTER* the entries already present in /etc/hosts.
### hadoop cluster
# format
# ip_addres
fully_qualified_hostname alias
10.1.1.101
hadoop1.lab.mycompna.com hadoop1
10.1.1.102
hadoop2.lab.mycompny.com hadoop2
# and so on….
Few things to note:
- the content of this file has to be distributed across the cluster
on all machines. DO NOT copy the file onto target machines, the
hadoop section needs to be APPENDED to /etc/hosts (see below for a
quick script to do it) - the first entry is IP address (usually an internal IP address)
- second entry is the FULLY QUALIFIED HOST NAME. This
makes sure reverse DNS lookup picks up the correct hostname - 3rd entry is a shorthand alias; It saves me some
typing. So I can just type ‘ssh hadoop1’ rather
than ‘ssh hadoop1.lab.mycompany.com’
One of common mistake that happens here is when host alias and fully
qualified hostnames are swapped.
following isn’t correct,
10.1.1.101 hadoop1
hadoop1.lab.mycompany.com
aliases should follow, fully qualified host names.
The hadoop cluster section of /etc/hosts file has to be distributed on
all cluster nodes.
How to distribute the DNS entries across the cluster?
There are confguration managemnet systems (CMS) like Chef and Puppet that makes distributing
config files on a cluster easy. For a very large cluster,
using a CMS could be a recommended choice.
Here is a quick way to distribute the files:
If you have SSH password-less login setup between master and slaves,
the following would work:
1) backup existing hosts file: (do this only ONCE!)
run the following script with ROOT privileges
2) save the hadoop specific DNS entries into a file
say ‘hadoop_hosts’ is our file that has the following content:
### hadoop cluster
10.1.1.101 hadoop1.lab.mycompna.com hadoop1
10.1.1.102 hadoop2.lab.mycompny.com hadoop2
# and so on….
3) run the following script; it will copy the custom hosts files to destination and append it to the existing /etc/hosts file
Checking DNS across the cluster
Here is a simple Java utility I wrote to verify that DNS is working ok in ALL cluster machines.
The tool is called : HADOOP-DNS-CHECKER , it is on GitHub.
Here are some features:
- It is written in Java, so it will resolve hostnames just like Hadoop / Hbase would (or at least close enough)
- It is written in pure Java, doesn’t use any third party libraries. So it is very easy to compile and run. If you are running Hadoop, you already have JDK installed anyway
- it does both IP lookup and reverse DNS lookup
- will also check if machine’s own hostname resolves correctly
- it can run on a single machine
- it can run on machines across cluster (as long as passsword-less ssh is enabled)
To run this, say from hadoop master:
- get the code (using git : git clone git@github.com:sujee/hadoop-dns-checker.git)
- compile: ./compile.sh it should create a jar file ‘a.jar’
- create a hosts file (‘my_hosts’) containing all machines in your hadoop cluster:
hadoop1.domain.com
hadoop2.domain.com
hadoop3.domain.com - first run this in a single machine mode:
./run.sh my_hosts
here is a sample output:==== Running on : c2107.pcs.hds.com/172.17.34.99 =====
# self check…
— host : c2107.pcs.hds.com
host lookup : success (172.17.34.99)
reverse lookup : success (c2107.pcs.hds.com)
is reachable : yes
# end self check— host : c2107.pcs.hds.com
host lookup : success (172.17.34.99)
reverse lookup : success (c2107.pcs.hds.com)
is reachable : yes— host : c2108.pcs.hds.com
host lookup : success (172.17.34.100)
reverse lookup : success (c2108.pcs.hds.com)
is reachable : yes - great. Now we can run this on cluster. It will login to each machine specified in ‘hosts’ file, and run this script.
./run-on-cluster.sh hosts2
if any error is encountered it will print out ‘*** FAIL *** ‘. So it is easy to spot any errors
Hope you find this useful; Please leave your comments below.

4 Comments:
By Mohammad Tariq 06 Aug 2012
Great post Sujee..quite often people struggle with DNS resolution problems as soon as they start their Hadoop journey..Would be very helpful
By Paul Baclace 16 Aug 2012
Thanks for the full treatment of this topic, Sujee. I thought I would not need this, but it seems to be an issue on HP Cloud at this time.
Do the names really need to be fully qualified? In the spirit of foo.local, I recently started using foo.wan and foo.vpn for some some setups.
By Sujee Maniyam 16 Aug 2012
Paul
no need for FULLY qualified domains. Just need to have ‘a domain’.
I’v used hostnames like ‘hadoop1.cluster’, as long as I had a matching entry /etc/hosts things are fine.
Amazon EC2 internal IPs have the same format ‘ip-1-2-3-4.internal’
By Akmal 04 Mar 2015
Thanks man, you’ve saved my day! Great explanation!