minimum temps for running old hardware?

On Mon, Jan 27, 2014 at 12:12 PM, Fred Cisin <cisin at xenosoft.com> wrote: In the machine rooms? Yes. At Pole, we use outside air controlled through themostatic louvers to cool equipment. By the time it's entering the room, it has been warmed up in the ducts to around -30F/-34C in the winter (maybe -10F/-23C in the summer). With the AMANDA experiment (~19KW in 15 racks), we had to balance the temps between too warm for the HV crates (the air ducts blew right on them) and too cold for the discriminator channels. We had to keep the room between about +41F/5C and +59F/15C. +68F/+20C was too warm. +32F/0C was too cold. The direction and intensity of the wind often caused larger-than-acceptable fluctuations in the quantity of outside air entering the intake making things tricky to manage when it would alternate between storms and calm. In 2005, a year between two of my winters, the environmental process controller in the Ice Cube Observatory building crashed when the louvers were shut. From the reports I got, the room warmed to +120F/+49C before anyone could walk out to the building to mitigate the problem. With +120F in the aisles, some of the racks were substantially warmer, locally. We lost several machines due to heat. When I returned a few weeks later, a couple of the lesser-used boxes were still broken, but all the critical systems had been returned to service. The flip side is, when it's +5F/-15C or +10F/-12C outside (which happens a couple of days every few summers), the air isn't cold enough to keep the room cool through the usual airflow, so we've had to turn equipment off or open doors to move more air. So... "free" Air Conditioning, but if the air stops or gets too warm, it all goes pear-shaped. -ethan