What thirteen pulled servers actually cost

Between 20260901 and 20260903, thirteen chassis of decommissioned datacenter hardware (three 1U, nine 2U twelve-bay, one 4U thirty-six-bay Supermicro units) became a working fleet, with Christoph on site and an agent doing the remote half. The sticker price of used enterprise gear is not the real cost. The rest of the bill arrived as parts and hours.

Counted over those three days: 3 of roughly 80 twelve-terabyte drives sold as refurbished and never run were dead on arrival, 1 of 3 RAID cards was dead, 5 of 13 CMOS cells (the small battery that holds a board’s clock and firmware settings) were flat, and one motherboard came with a riser slot that had been physically damaged before it ever arrived. None of that is a defect rate you would accept from new hardware, and all of it is normal for pulls.

The second half of the bill is tooling that new gear never needs, because new gear does not have to distrust itself: a survey script to find out what each chassis actually contains, a routine to clear a previous owner’s RAID configuration, and a verify script that runs after any chassis is touched. The verify script checks drive counts, NVMe function counts, an LLDP neighbour on every network port (LLDP is the protocol by which a switch and a host announce themselves to each other), bonds up, no failed systemd units, and the CMOS battery voltage. Manual BIOS passes were replaced by a gold settings file applied as code.

The habit underneath all of it is the one that carried through every later hardware lesson here: the definition of done is a clean print from a check, not a person’s memory of having looked.