Re: IOC, QServ, and AMSG / alarm.message

[Ralph Lange via Tech-talk <tech-talk at aps.anl.gov>]

I completely agree with Michael's statements on the philosophy of alarms. (Careful - that BEAST paper is historical and not the current Phoebus alarm system.)

At nuclear facilities, there's an even stronger operational idea of what an alarm is...

Operations expects a maximum of a few hundred alarms for a facility like ITER. With less than 10 of the ~170 systems integrated, the control system already has way over 100k EPICS alarms... Oops.

Just as Michael said: as a consequence, EPICS alarms are downgraded to engineering hints, keeping the worlds apart. Some of them will be inputs to whatever Operations will use for alarms.

In the past (no PVA or AMSG involved), I often worked with DESC and the status strings by asking for "<DESC>: <stringVAL>" to be a meaningful description of the status/alarm. For the mentioned example, something like

record(bi, "$(P):T3TP4_FAULT:STATUS")

{

      field(DESC, " INJVAC – T3TP4 Status")

      field(ZNAM, "Fault")

      field(ONAM, "Ok")

      field(ZSV, "MAJOR")

field(INP, {pva:{pv:"$(P):T3TP4_FAULT:STATUS_C", proc:true}})

}

Doesn't always work, but creates reasonable strings most of the time.

Cheers,
~Ralph