I am testing out ganglia for an application here at LANL. I have a two
node test cluster, and both nodes are gexec capable, and gexec seems to
work on both. The notes in the documentaiton on using a dynamic node
list indicated that load balancing is done based on 1 min load average.
If I use the first node (.100) as the value for GEXEC_GMOND_SVRS,
executing something like
'gexec -n 1 -d nbench ' always runs on the other node (.101) no matter
what it's load is. will stubbornly start a half dozen nbench benchmarks
on the .101 node and never start one on the .100 or unloaded node.
executing a benchmark cluster wide with 'gexec -n 0 -d nbench' will
start an instance of the benchmark on both machines, regardless of load
of each machine.
There seems to be no load balancing going on at all.
Is this correct behaviour, or am I missing something?
thanks!
r.
--
/**************************************************************************/
Russ Gritzo email: [EMAIL PROTECTED]
"I haven't lost my mind -- it's backed up on tape somewhere."
/**************************************************************************/
