Improvement #1519
Improve Map/Reduce performance by not stopping Z2
Start date:
18.09.2013
Due date:
% Done:
100%
Estimated time:
origin:
Description
Currently map-reduce jobs running with Z2 will stop Z2 after the task. In case the process is configured to be re-used this is actually hurting performance and something is in the end always leaking classloader-wise. Therefore we should add an option to job and general config to simply leave z2 running until the current process is being actively stopped by hadoop.