Re,
sur le load average
vu dans le code de solaris
Note that this is just a heuristic;
388 * if we take multiple laps though clock() without
returning from
389 * the interrupt because we have another clock tick
pending, then
390 * the pinned interrupt could be released by one of
the previous
391 * laps. The only consequence is that the CPU will
be counted as
392 * in idle (or wait) state once the pinned interrupt
is released.
393 * Since this accounting is inaccurate by nature,
this isn't a big
394 * deal
et la meilleur
/*
842 * End of precision kernel-code fragment
843 *
844 * The section below should be modified if
we are planning
845 * to use NTP for synchronization.
846 *
_______________________________________________
Solaris_fr liste de diffusion en français pour Solaris, sur toutes architectures
[email protected]
http://x86.sun.com/mailman/listinfo/solaris_fr
