Thursday, February 25, 2010
UPS: The Autopsy
I performed a few more experiments with the dead UPS and noted that after it had sat for a few days I could get four or five software self-tests out of it as long as there was zero load on the battery.
Then it dawned on me that I didn't have a dead UPS. I had a dead battery.
After that I decided to put its dead carcass on the slab and rip its guts out.
I was quite surprised to find that its insides were spotless. For having rested for four years in the same spot I was astonished that it wasn't full of cat hair, parakeet feathers, dust mites, and all the other detritus that collects inside the many computers hooked up to DinkNet.
Four years! I have to clean out the computers every six months or the fans get clogged!
I pulled the battery out, which was a much more difficult operation than it should have been. The two APC UPS's I have both have easy-access battery compartments, but this thing required a near complete disassembly to remove the battery.
Once I got the battery (actually two batteries taped together and wired in series) out, I pulled the part number (CB1270 for those who care) and looked it up on the 'Net. An equivalent replacement was forty bucks and the manufacturer clearly stated it would last three to five years.
So... there was never anything wrong with the UPS at all and four years is about what you'd expect to get out of a battery anyway.
Anyway, I ordered a replacement battery. This UPS will be dedicated to keeping the Proxy Project up and running, since that particular box runs 24x7 and takes a lot of abuse whenever the power blinks out. Its an XP box, my favorite UT99 platform, but its primary purpose is to run a Linux VM for proxy testing and database operations. Since I despise the Windows software Powercom distributes with these things, I'll probably run NUT on the VM to monitor it and WinNUT on XP for shutting down the physical box.
That's really an odd way of doing power management, but it's the Hinky Way!