hama-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Behroz Sikander <behro...@gmail.com>
Subject General questions
Date Fri, 28 Aug 2015 02:42:32 GMT
Since, I am pretty new to cluster configurations, I need some suggestions
on how to solve below mentioned problems efficiently.

*1- Password-Less SSH between machines of cluster*
To have a working cluster, we need password-less SSH access between all the
machines in cluster. Till now, I was manually doing them because I had only
3 machines. I am moving to 20 machines now. So, it is a lot of work. How
teams who manage 100s of servers solve this problem ? Bash scripts ?

*2- Dynamic IPs*
This is the biggest problem. Every time I restart my cluster, all the
machines get new IPs. It means that I need to modify my /etc/hosts files on
all machines. Also I need to verify my password-less SSH logins. Getting a
static IP is difficult in my current setup. So, how people solve this
problem ?

*3- HDFS data*
As per my current understanding, HDFS has namenodes and datanodes.
Namenodes contain the all the information about chunks and where they are
placed. Now, lets assume I restarted my cluster and got new IPs. My whole
HDFS data will be messed up. Again how to solve this problem ?

*4- Monitoring*
Hama provides a Web GUI to check the basic information about the job. But a
few things seem to be missing like bandwidth, cpu and memory usage on
cluster and individual machine level.  Are there any third party tools that
can be integrated in cluster to monitor Hama ? (AMBARI maybe ?)


  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message