eurotherm: handle missing sensor better

[FreddieAkeroyd]

As a developer i would like the eurotherm to handle a missing sensor better. As seen recently, the IOC was configured for sensors 1,2,3 but 1 had been removed. This resulted in calls to read sensor 1 timing out, but the queue of such reads caused other reads on 2 and 3 to time out, though occasionally a read from 2 or 3 would get through, so they were intermittently showing the right value but mostly in alarm.

maybe the read chain could be modified so that if the first read fails it does not continue the others in that chain on that iteration. It looks like may initiate reads viarecord(seq, "$(P)READ") ` every second. given timeouts in reads, i feel one approach may be to

• separate out one particular read (may even be a duplicate of something we do) to a separate read loop at say 5 second interval. if this read fails, the record will go into alarm
• change record(seq, "$(P)READ_RESTART")` to be a calcout or similar sort of record. what we want it to do is scan every second, but part of this is to check the alarm status of the 5 second scan record above and if that is in alarm not to initiate the record(seq, "$(P)READ") sequence.

Acceptance criteria

communication with a missing sensor does not affect other working sensors

Time in Planning

1:00:39 02/05/24

[LowriJenkins]

Freddie plans to make this ticket clearer, if you don't understand it, ask him.

[isaachilly]

Rework info

The Lewis test need to simulate just one addr being disconnected not all of them which it is currently doing. This may need writing a new Lewis command that only simulates the disconnection of one addr. Test needs to check it can write and read values from the other addr/s with 1/more addr being disconnected.