Hi, We recently implemented Ganglia on a large scale cluster (2500+ nodes) and found out that we have hit a limit in the memory summing. As soon as we hit the 4TB limit, the counter reseted to 0 and started summing again. This is not that important as we now read it modulo 4TB but how about, in the future releases, storing that kind of info in MB (I think it is currently stored in KB) which will move the limit to 4EB?
Just my 2 cents... Regards, Laurent
