[hinkley]

That user posted an update:

"everything minus core network gear was manually being unplugged from any PDUs to help the control the initial power-on."

Can't we... Why don't we have rack hardware that can handle this situation? I thought some HDD RAID solutions had circuitry to keep them from browning themselves out while spinning up the disks. I guess I'm surprised this isn't a solved problem at the rack level now.

Or have we been so focused on never cold booting a rack of servers that we haven't spent any effort on foolproofing of cold booting a rack of servers?

[Edit: answering my own question] apparently these exist and are called Managed PDUs. Can we deduce WF doesn't have them?

[mustardo]

Some servers "stage" power on to disks. as spinning up disks is often the largest source of inrush current for e.g. a 10 disk server might only apply power to bays 2 at a time with a delay between each, this doesn't solve the inrush current problem of switching on a rack full or servers together (where managed PDUs come in)

[rolph]

it would seem to be a possibility. im also thinking 10 finger power up sequence, as in: insert power cord 1 into powersocket 1, wait till audible beep, insert power cord 2 into power socket 2, wait till audible beep etc. :-D