Schlagwort-Archive: hadoop

anotherServerInstance

as a friend from a provider in berlin gave me the access to a vhost he had, i received another donation from him. a vhost instances with some more ram, as i had on the machine since now was now in the park.

so i had three machines to build my cloud further. since 2010 i was working with big-data tools and the hadoop-framework, a distributed file-system was the main aspect to build such thing. with the new server, i was able to raise the replication from 1 to 2. that was not the replication-factor of 3, which could be seen as minimum for hot-fail-over redundancy systems, but on the right way.

i introduced the machine to the cluster and configured the redirect-proxy the same way, as the machine in jena. after some tests, i had failures with „to many open files“ and i noticed, that the vhost-guest-system was heated to much with the hdfs, so i configured them further.

beside the cluster, i was able to use them as streaming-server for the radio as well, as i switched chromanova.fm on, with two channels as well. that was nice, to have capacity in bandwith that way and the scale was nearly linear to the count of machines.

now i was able to play harder with the layer between the tomcat-servlet and the byte-stream, given by the filesystem. the new machine had 4 gigabytes of ram and i gained the headroom for the caching on tomcat much higher then possible before.