gladky
commented
6 years ago Ideas of Hannes:
As Andre also remarked yesterday, we need to keep the history of what was shown to the shifter. A few ideas:
When you update numbers such as dead time / trigger rate we should have the min, max and average number displayed in addition to the current number.
When the subsystem or partition of a problem changes we should break the condition and keep the old one in the history so that we have a history of what was shown to the shifter.
When FEDs that are responsible for a condition change, we should show the FEDs that currently cause the problem with and in addition the FEDs that previously cause d the problem (maybe greyed out, or the current one in bold). We could take a similar approach for when the partition changes.
It might be a good idea to show the updates to a condition in the event log (on the right hand side) as update events. With that we would have the complete history of what was shown to the shifter at any given time. But of course this would flood the log.
Problem:
There was a dead time from ECAL fed 602 at this timestamp: http://daq-expert.cms/daq2view-react/index.html?etup=cdaq&time=2017%2008%2030%2016:19:56
In the next snapshot there is no dead time from 602, there is from 582/583. The condition was satisfied first by FED 602, later it was hold by other FEDs. But the notification was based on first snapshot.
Desired solution: A. Yield separate condition for each component. It might be good idea unless there is a lot of individual conditions generated and the browser and dashboard become buried with the same type of condition for different subsystem/feds. B. State the fact that condition is now held by different components in one condition. Introduce some kind of changeset.