the tools that coordinate went away

A summary in this record opens with an outage: the code host and the shared task board were both unavailable at the same time. The same sentence carries the second half — local state was intact — and the two halves belong together, because jointly they say which layer failed and which did not.

That distinction is the whole incident. Coordination substrate and durable state are different things, and most operations discover they have quietly conflated them only while an outage is happening. Here the coordinating layer could disappear without taking the work with it, because no lane's state existed only on the board or only on the host. It existed in per-lane records that had been reconciled continuously for hours. The shared substrate was a convenience, not the place the operation lived.

The cost is structural and it is paid every cycle rather than once. Designing for this makes the reconciliation work the durability plan, so it is expensive on the days when nothing fails, and there is no visible reward for it. It also means the outage does not appear in the record as a recovery story: nothing had to be rescued, so the entry is one line about tooling. An operation that judged itself by the drama of its incident write-ups would score this badly.

What it proves is that the line between coordination and state was drawn deliberately enough that losing the coordinating machinery cost time and nothing else.