> -----Original Message-----
> From: Testbed Support for GridPP member institutes [mailto:TB-
> [log in to unmask]] On Behalf Of Stephen Jones
>
> I don't know what the big deal is - put it in a DB, write some updater
> scripts, make an XML extract every hour by cron and send it to whoever
> subscribes to the feed.
>
Well, there are several strata of problem. There are the trivial details like the schema not being able to tell the difference between hyperthreading units and actual cores (they're both 'logical CPUs'), there are the major implementation problems like the need to advertise separate subclusters for each generation of nodes, and there are the fundamental conceptual problems that the whole principle doesn't work at all for non-traditional 'cloudy' resources where the total number of nodes and the details of the hardware may not even be known.
But, the practical point here and now is that these figures are already wrong and this request is to make them still be wrong, but in a cosmetically different manner. Or we could just stop raising alarms and tickets about this since no-one actually cares about these figures being correct, because they're not anyway.
Ewan
|