
Each sled has two of the now super-cheap Xeon E5-2670 CPUs, 8x8GB RAM, and an eBay sourced Samsung 480GB Enterprise SSD drive. 480GB is too much, but they were dirt cheap and I'm planning on the extra capacity improving endurance since these servers hit the disk moderately hard. These servers run CPU-heavy but network-light tasks, so I removed the 10G Ethernet Ethernet mezzanine cards, saving some 550 watts per rack.I am so jealous... so one CPU each server and 64GB of memory?
That's my Danish Ops Director there. Probably holding his nose to ward off the smell of the quirky OCP PXE behavior.@dba looking at that individual sitting next to the racks made me think Seinfeld a bit - is that a pick or a scratch?
Very cool pics. OH?
Thank you for advices and I'll try themHow's your CentOS 7, will you be able to halt/reboot it? I have 8 of them connected to ACUPWR 1500w, all of them cannot reboot/halt correctly, I have to reset/power off them manually.
Yeah..that was a pretty bad stench.....That's my Danish Ops Director there. Probably holding his nose to ward off the smell of the quirky OCP PXE behavior.
What model card worked for you? I know you were saying that some newer cards are too large and/or not compatible.Yeah..that was a pretty bad stench.....
Lesson 1 when working with these servers, buy a low end graphics card so you can see whats going on during boot and install.
Yes PXE is the default boot unless changed. If you find a GPU to put on it, it acts like any other system and you can get to the BIOS that way.1) does the BIOS support PXE boot? Can you enter "enough" of the BIOS to do it without the debug kit?