Home
| Calendar
| Mail Lists
| List Archives
| Desktop SIG
| Hardware Hacking SIG
Wiki | Flickr | PicasaWeb | Video | Maps & Directions | Installfests | Keysignings Linux Cafe | Meeting Notes | Linux Links | Bling | About BLU |
Indeed. Throttling down the server power usage is exactly the purpose of my exercise here. I want to show mostly the availability of this solution rather than any determination of the utility of same. According to http://ebscosustainability.files.wordpress.com/2010/06/data-center-energy-efficiency.pdf even an efficient server still uses about half its full power when doing virtually no work so it would be great if we had an easy to use, CPU-load checking, aggressive power management system. I have tested using cpufreq-set to slow my boxes down to a crawl with very little effect on power usage so I think using suspend or hibernate are a couple of the few options I have left. I'll try some of your suggestions; hooking up a digiboard and multiplexing out serial access, or using jabr's idea, but I wonder if suspend will turn off access to everything but usb and / or ps2, which is where it is expecting to get a mouse wiggle from... Thanks for the help... I'll keep you posted... I also bought some IP-addressable plugstrips and will test upsd and nut with it. Shutdowns may be a little too aggressive though; I don't think I have that much time between jobs! > On Thu, May 17, 2012 at 09:57:53AM -0400, Kurt Keville wrote: >> Thanks... I'll give that a test... my big problem is (or will be) >> lack of physical access to the servers and I figure this has to have >> been bumped into out there in Datacenterland by someone... Federico >> pointed me at powernap which appears to have some hook options that >> might fit the criteria... > > Another potential solution to this, depending on the details of your > arrangement, is to set up console access on the serial port, and then > use a terminal server or similar device to access the console over the > serial port. Especially if you're going to have a rack full of > servers set up this way, it can come in quite handy. > > http://www.howtoforge.com/setting_up_a_serial_console > > IIRC server-class hardware (maybe all hardware these days?) can also > be configured to provide bios access on a serial port, e.g.: > > > http://h20000.www2.hp.com/bc/docs/support/SupportManual/c00440332/c00440332.pdf > > I agree with Jerry that your server machines should have ACPI power > settings disabled so they just never sleep. Much modern hardware does > have a wake-on-lan feature, though IIRC you need to send it a > particular type of network message for that to work (and it needs to > be enabled). But there's really no reason for a machine intended to > act as a server to ever go to sleep, unless you are the only one who > will ever access it, and you're prepared to wake it up every time you > want to do so. > > -- > Derek D. Martin http://www.pizzashack.org/ GPG Key ID: 0xDFBEAD02 > -=-=-=-=- > This message is posted from an invalid address. Replying to it will > result in > undeliverable mail due to spam prevention. Sorry for the inconvenience. > >
BLU is a member of BostonUserGroups | |
We also thank MIT for the use of their facilities. |