Using 0.90 RC3 on our 7 node clsuter, the first time (after cluster start) our map tasks ran quite fast writing to hbase (client side buffer size being 17MB).
If I run the same hadoop job soon after the first finishes, there is noticeable slow down. Connections to zookeepr quorum wasn't high - 71 including connections from region servers. On region server(s), jps didn't show map task processes although ps could. jstack reported: 13331: Unable to open socket file: target process not responding or HotSpot VM not loaded The -F option can be used when the target process is not responding Advice for further debugging is welcome.
