[Olympus] SlowControl

[Anton Izotov]

Dear Jan,

IMHO. Except HV and even with HV SC data stream is very 
small in comparison with main data. So I would propose to write everything 
to DAQ user events as well as to SC db. Just to be on a safe site.
However, data analysis will go slower if one has to dig  SC data out from 
huge amount of DAQ data. It is much easier to take it from one place.

So we can store everythin in CS db and in DAQ. But use DAQ stored data 
only if it somehow lost in SC db.

Anton


> 
> Dear colleagues,
> 
>  in Mainz, the slow control saves everything (event driven) in its own
> database, but at the start of a run, the complete status quo is dumped
> into the data acquisition file. That is about 1 to 3 MB of data, so no
> problem. It is so much easier to analyse your data if you have
> everything consistent in a single location, and you don't have to match
> and mix.
> 
> We even mix important slow control information into the readout event
> stream, like target parameters (pressure, temperature), event scalers
> etc. Of course not for every event, but on every change of these values.
> That proved *very* useful.
> 
> Redundancy in the data saved my analysis several times. I suggest we
> store the start and stop of each run in the slow control log, and for
> every event as much timing info as possible (run time, live time, dead
> time from scalers, unix time). Do we have a plan for the luminosity and
> dead time estimation?
> 
> Best regards,
> Jan
> 
> Alexander Winnebeck wrote:
> 
> > @ John: I wonder if you mix up two things here. There is the slowcontrol 
> > and the online monitoring. The slowcontrol will control the HVs, 
> > temperatures, flows, machine parameters etc. and the online monitoring 
> > will evaluate the incoming data from the detectors, i.e. showing TDC and 
> > ADC spectra, count rates, hit distributions etc.
> > To have an offline matching of sc data and the data of the detectors, we 
> > might have a table in the sc with the time of begin and end of run with 
> > the run number, or the unix time is stored in every event in the data 
> > stream.
> > 
> 
> 
> -- 
> Jan C. Bernauer                       email: bernauer at kph.uni-mainz.de
> Institut f?r Kernphysik               tel  : 06131 - 39 25826
> Johannes-Gutenberg-Universit?t Mainz  fax  : 06131 - 39 22964
> Johann-Joachim-Becher-Weg 45
> 55099 Mainz
>