Recently I have been working with Dell’s XS23-SB server, it is similar to a Poweredge C6100 in that it has four blades in a 2U Chassis. It has 12 3.5 inch hot swap drive bays in the front and the four blades are accessible from the rear with VGA, USB and networking ports.
The XS23-SB is part of Dell’s DataCenter Solutions (DCS) department so they are put together with commodity hardware. The motherboard is a Super Micro X7DWT which supports dual 5400′s, I went with dual Intel Xeon L5420 CPUs. It has eight FB-DIMM ram slots and can support up to 64GB per node. There are four Sata ports on the motherboard that connect to the drive bays in the front via the backplane.
Dell XS23-SB Disassembled
Below you can see the XS23-SB disassembled. The four nodes are in a 2×2 orientation in the chassis. The top two are on a shelf which you remove with little difficulty after unhooking some of the cables. In the photo you can see two Xeon L5420 CPUs waiting to be installed.
XS23-SB Top shelf with two nodes
Below you can see the top shelf removed. This photo shows two identical motherboards side by side sharing the same power supply. They each take two CPUs and have their own VGA / USB outputs in the back of the chassis.
While setting this server up I ran into a unique problem. Using identical hard drives some of the nodes recognized some of the drives, and some of the drives simply wouldn’t work. I tried troubleshooting every combination of internals and inspecting the backplane, checking connections etc. I troubleshot the drives one at a time in ever node, no luck. I assumed at first that I had blown the drives due to static shock even though I usually wear a static strap.
I am working with two servers and I wanted to get all eight nodes up and running so I cannibalized a GDrive RAID-0 enclosure in order to use the two drives as OS drives, since at the time I was assuming that I had a handful of dead hard drives. I installed the first hard drive into a node and it worked fine. The second one however did not work.
This was a fairly frustrating experience and I had to regroup my thoughts. After some deep reflection and reviewing the process I had taken I theorized that since I was using drives that had been in raid configurations there must have been raid data on some of the drives that the X7DWT SATA controllers did not like.
The Solution
After doing a low level wipe the newly zeroed drives that did not work before, worked like a charm. It was as simple as wiping old RAID data from them!
