fix(esh): revert the forced 10G — the link was flaking hard, operator called it

Both ends are back on autoneg at 1000 Mbps and stable; UDM SFP+2 restored to
autoneg True / setting_preference auto from the payload staged before the
original change. ~50 s of renegotiation, nothing rebooted.

The force failed progressively and I called it too early. Errors on the
ESH-Media side went 200 at link-up, 221 at 42 minutes, 416 by four hours, with
flapping the operator noticed before my soak had accumulated enough to raise
it. I read a two-minute flat window as a plateau and said so; a marginal link
declares itself over hours, not minutes.

Do not re-force this port. The cable is being replaced at the weekend and a
correctly-coded DAC negotiates 10G on its own.

docs/runbooks/esh-media-dac-10g.md

@@ -3,30 +3,27 @@
 2026-09-04. The link runs at 10G by an explicit rate override on the UDM side.
 **Read the pinned-port warning below before touching that port.**
 
-> ## ⚠ CURRENT STATE: forced 10G, KNOWN MARGINAL, held deliberately
+> ## CURRENT STATE: REVERTED to autoneg / 1G. The force is gone.
 >
-> **Operator decision 2026-09-04: leave the override in place.** He has a
+> **2026-09-04 14:38 — the operator reported the 10G link "flaking hard" and I
-> replacement cable and will fish it through the garage at the weekend. Do not
+> reverted it.** Both ends are back at 1000 Mbps, UDM SFP+2 is `autoneg True /
-> revert it and do not re-diagnose it — this is a held position, not an
+> setting_preference auto`, ESH-Media is reachable and stable. The revert took
-> oversight.
+> about 50 s of renegotiation and nothing rebooted.
 >
-> The link is **not clean at 10G**: errors accumulate steadily (200 → 221 rx over
+> **The forced-10G experiment failed, and the failure was progressive.** Error
-> 42 min, ~1 per 23 MB, roughly one bad frame in 15,000) and it dropped once, at
+> counts on the ESH-Media side over the life of the force:
-> 10:12, recovering on its own. Nothing rebooted. TCP rides over it; expect
-> retransmits and jitter rather than visible corruption.
 >
-> When the cable is swapped: **remove the override** and let autoneg work. A
+>     10:03   200      link-up burst
-> properly-coded DAC negotiates 10G on its own — that is the whole point.
+>     10:48   221      +21 in 42 min  -- the first sign it was not plateauing
-
+>     14:38   416      +195 over the next 4 h, plus user-visible flapping
-    link      ESH-Media SFP+1 (port 11)  <--DAC-->  ESH-UDMPM SFP+2 (port 11, eth10)
+>
-    cable     OEM "SFP-H10GB-CU3M", serial CSC200803000004, 3 m passive twinax
+> ⚠ **The lesson, and it is mine to own: I read a two-minute flat window as a
-    before    both ends 1000 Mbps, autoneg on
+> plateau and reported it as one.** The counter simply had not moved yet. A
-    after     both ends 10000 Mbps
+> marginal link declares itself over hours, not minutes — and the operator felt
-
+> the flapping before the soak I left running had accumulated enough to call it.
-This link backhauls the USW Flex XG "ESH-Office-Vuong", the E7-ESH-Media access
+>
-point, and the Zigbee coordinator — the office switch and the AP both connect at
+> **Do not re-force this port.** The cable is being replaced at the weekend; a
-10G and were squeezing through a 1G uplink, so this was a real bottleneck rather
+> correctly-coded DAC negotiates 10G on its own, which is the actual fix.
-than a cosmetic one.
 
 ## The diagnosis, and what made it certain
 
@@ -78,12 +75,12 @@ pre-change `port_overrides` array was written to a revert payload before the
 forward one was sent. If a future attempt needs *both* ends forced, that is a
 different risk class and needs the operator's sign-off, not an agent's.
 
-## ⚠⚠ THAT PORT IS NOW PINNED — a 1G module will not link
+## The port is NOT pinned any more — the override was removed 2026-09-04
 
-UDM SFP+2 is `autoneg False, speed 10000`. Plug anything that is not a 10G
+While the force was in place, UDM SFP+2 was `autoneg False, speed 10000` and a
-module into it and the link stays down with no obvious cause. If the DAC is ever
+non-10G module would not have linked there. That is no longer true: the override
-replaced with a properly-coded one, **remove the override** and let autoneg do
+is gone and the port is back on autoneg. Recorded because the warning stood in
-its job rather than leaving a pin nobody remembers.
+this runbook for four hours and someone may remember it.
 
 ## What the link bounce cost