-
Type:
Task
-
Status: Done (View Workflow)
-
Priority:
Normal
-
Resolution: Done
-
Component/s: ics_alertsActor
-
Labels:
-
Story Points:1
-
Sprint:PreRun22May2025
As reported on slack in ics channel :
Looking at the alertsActor yesterday during the shutdown, I realize something might be counter-intuitive, at least it's confusing me.
?? In case of timeout, it's sort of a special case because this happen when the keyword are not being generated.??
?? But as you know when alertsActor broadcast values to STS it's always a timestamp and a pair of value, text.??
?? If a normal case, keyword is being updated : timestamp=keyword.timestamp, value=keyword.value and text="OK" or something else depending on the logic/value??
?? In case of a timeout, keyword is not being updated, so the timestamp become 'now', the value=last valid value, and the text is "NO DATA since ... ".??
?? I realize it might be a problem on STS side because it means that we're pushing the last valid value with the current timestamp.??
?? You could say, (if there is a way), that's fine we can just filter the values where the text is "NO DATA since.." because we know those are incorrect.??
?? But in the case of xcu, when the mode is set to offline, then the message become "OK", because we ignore the timeout on purpose in that case, so that logic would not work.??
a simple solution would be not to push the last valid value but use the same constant we use for FITS file which are defined in ics_utils :
# Defining EXPIRED and INVALID in only one place. EXPIRED = -9998 INVALID = -9999