LB-2207: load balance survey, fen, rows 4 and 9 — the meters agree

Work order LB-2207 (annual load balance, rows 4 and 9, 2026) · Site fen · Surveyor one, sub, 14 years on tools, no site system rights · Method PDU per-outlet registers read out, then each circuit clamped at the bar and the two sets compared. The clamp is the instrument. The register is paperwork, same as any paperwork.

I do not do software. What follows is what the meters say, and I have checked that the sentence before it is the only claim I make.

Circuit 9-4/3 (outlet U21)

The outlet for U21, rack 9-4, reads 0.18 A at 230 V. That is forty-one watts, and that is the whole reading, so it is the whole page.

I clamped the circuit at the bar at 10:40, with the outlet switched off at the PDU and the breaker for the circuit pulled at the board at 10:20 by my own hand, dead all day, nothing to measure. The clamp says 0.18 A. The register says 0.18 A at the same minute. I pulled the fuse out of the clamp to prove the clamp to myself, and the clamp read zero with the fuse out, which is what a clamp does.

U21 is where hesper-01 stood until August, when a crew took it out on (DCM-1188). The circuit was switched off when the node came out. The only work order since was a blank panel, fitted in September (DCM-1188-B). I asked for the register back to 2021 to see where the forty-one watts starts. It starts 2021-04-06 and never moves. Not four amps of inrush, not a power-factor argument, not capacitor bleed: bleed droops, and 0.18 A is flat to the last decimal the register holds, every day since. A server on but not doing anything draws about forty watts.

Circuit 4-2/2 (corvid-b)

The cooling failure in August tripped two of the config-store pair (PIR-0918). Facilities re-terminated circuit 4-2/2 after the failure, and while the panel was open the circuit was off: corvid-b powered down hard at 04:02:28 on 2026-08-11. The outlet register holds the server's normal 1.9 A through the whole window, and the clamp agreed when I took it at 11:15 on survey day, four days after the box came back, which proves nothing about that night except that the meters did not move when the server did.

The rack total on the incoming equals the sum of the outlets, and there are two servers racked in that rack. The arithmetic is right. Two boxes, three loads. The arithmetic not lying is not the same as the arithmetic meaning nothing.

Notes

Stores, 2026-08-21. The old blank from 9-4 sat at stores awaiting return while I waited on a van. Parcel scale is for shipping mass and I used it for shipping mass: 1.1 kg over the nominal mass of a 2U blank, and I lifted it once, which was enough to tell me where the kilogram is. The crew who fitted the new one noticed it by hand the same week. I photographed the scale. Pictures are readings.

U22 field end. There is a pair punched down that the cross-connect sheet does not have. A pair with no circuit number is not a circuit, the sheet is somebody else's trade, and I said so and left it.

I am filing against the work order, which asked whether the rows balance. They balance. Every meter agrees with every other meter, and that is the finding, and no amount of asking me what I think changes it.

— lb-crew-9, sub, sticker on the clamp case

LB-2207: the meters agree