Hi Everyone!
Just had the switches firmware upgraded. Lamboot still boots very slow
(maybe). For 84 P4/2.4 GHz nodes being booted over a gigabit LAN, it
takes approximately 20 minutes. Is this normal or much too slow?
Another question: since the resource of 84 CPU's in 20 minutes isn't
exactly insignificant - could one instead lamboot the whole cluster (210
nodes) once and let PBSpro send the mpirun part, i.e. the main job to a
selection of nodes? (ssh or tm?) Would this at all be possible? If this
is possible, are there any objecting arguments?
Best regards, Jess Michelsen
|
