By Michael Oliver -
I have been toying with utilizing WOL in conjunction with our render farm / workstations. ?I have been monitoring usage and it seems that we often have machines sitting idle over the weekend for a period of time after heavy renders complete. ?The idea is this:
Implement a policy where machines sleep/turnoff/hibernate after a set time period. ?
If the render farm manager notices it needs a certain number of machines, it pauses the job, sends WOL magic packets to the machines it needs, and then resumes the job once those machines come online. ?Once the render is complete and the period of time elapses again the machines go back to their sleep/off/hibernate state.
Anyone else play around with this? ?Can you think of any cons?
I know this would mess with the uptime stats everyone brags about but I would rather brag about my power bill ;)