I am planning to start a small in-house data center for my company. We will be putting around 10 rack servers in the basement. However, our office and staffs are located on the 5th floor of the building.
I am trying to brainstorm an idea with experienced data center professionals regarding remotely hard resetting the servers. Lets say one of our servers hanged or for whatever reason I need to hard reset the server, is there any possible way to do that remotely? In the perfect world, if the server hangs then it will just auto reset the server. But we don’t live in the perfect world. I thought about setting up a monitoring server or a script that will restart the server but if the server is hanged and it is not responding to any outside command or input (including LAN), that is not possible.
I would really like to know if there is any other way to hard reset a server (or any computer really) without going down to the server room and pressing the reset button manually. I bet this is not the first time someone thought about this irritating problem and there must be a solution out there. Are you guys using any hardware or software solution to address this kind of issues?
If you want it fully automated you can install a watchdog card in each server. You install a card, PCI, PCIe, or ISA, and a small software program. If the software stops communicating with the card, then the card will hard reset the server.
Did you find anything yet? Typically you need something like APC Switched Rack PDU which will cost you several hundred dollars to a thousand+.. or you can try PwrUSB's Watchdog version for less than a hundred. If I am recalling right, I saw those going at at $89 something on Amazon. Or try their website - pwrusb dot com
With PwrUSB you don't have to worry about network connectivity at all. PwrUSB will be attached to your server or computer locally thru USB. It comes with a software bundle that will monitor and send something they call "heartbeats" to the PwrUSB Power strip. IF your server hangs, the heartbeats will stop and after a predefined time, the power strip will auto hard reboot the server! Neat right?