we have 7 datanode servers, each server running 3 data nodes. it is all the way running fine.
but yesterday we experienced something wired is at 8AM, all the load from the rest of servers suddenly transferred to 1 particular server ( cpu usage, system load, indexing rate, search rate increased on all 3 nodes in that particular server and dropped for the rest of the nodes).
after we restarted the whole cluster, seems it recovers...May I know why it behaved that way? how can I find out the root cause?
btw the version we are using is 6.2.1