I compared our FR10 fixture-check ledger with the physical housing count after a monitor outage. The ledger gained one completed item on reconnect. Nothing moved during that interval.
The PLC retains the completion flag and job register. Our monitor treats the first true flag after connection as a fresh event. I can explain the extra row, but I am unsure how to reconcile it when an outage might also hide a genuinely new completion.
An attempt number. It increments when the cell accepts another fixture check. The recipe has its own register. The duplicate rows have the same attempt number.
I've been looking at restart duplicates on another reporting job. Saving the attempt with the result is important, but check whether your number can start again after a controller restart. Same number might not always mean same attempt.
Keep the duplicate row visible as a correction, please. Quality may already have seen that total. Quietly deleting it makes the report appear to have always been right.
Daniel, yes for the correction history. The new importer should still refuse to count the same identified completion twice. We don't need a new production item every time a connection wakes up.
The attempt counter survives a connection loss. I have not established what happens after a controller restart or replacement. That is now a question for the controls owner.
That's the other hole. If only the latest attempt is retained, Ella might reconnect to a perfectly real completion and still have missed several before it.
Mina's question needs an answer before choosing a fix. Ella, your duplicate example can test one part of it now: reconnect repeatedly with those same retained values and check that the count stays put.
Yes, multiple checks can finish during an outage. The current monitor only reads the latest retained result. We have reproduced the duplicate from a saved sequence without involving the cell, and the reporting change counts that sequence once.
Good, but don't fill the missing interval by subtracting attempt numbers. A number allocated to an attempt isn't automatically a completed housing. I nearly suggested that before remembering your counter advances on acceptance.
We will mark that interval incomplete. The cell owner is checking what history is available to recover individual outcomes. Counter restart behaviour remains open as well.
Who can close that incomplete interval in the reporting tool? I'd want the reviewed outcomes linked to the adjustment, with the earlier total still explainable.
Quality owns that adjustment. We have agreed the duplicate correction wording, but not implemented a way to reconcile missing outcomes. The dashboard change is still a test version.