Problems with GlueX runs in RCDB

[aaust]

There seem to be broken links in the rcdb entries for runs 30787 and 30788, which crash our simulation software:

JANA ERROR>>File with name "/home/hdops/CDAQ/daq_dev_v0.31/daq/config/hd_all/LIST" doesn't exist for run 30788

[nsjarvis]

Hi Dmitry,

at present runs 30344 30390 30787 30788 are unusable in analysis because the MC contains 0 trigger bits. Our code does not crash these days, it just checks the trigger bit for each event and produces nothing if there is no trigger bit.

I can see that the run config file is missing from RCDB for 30344 and 30787 and 30788  (but it's present for 30390 so Idk what is going wrong for that run, maybe it got corrupted).

What do we need to do to fix this? Should we upload the files again? How? Please could you provide instructions?

Thanks for your help,
Naomi.

[DraTeots]

it just checks the trigger bit for each event and produces nothing if there is no trigger bit

Could you please clarify what are those trigger bits and where do they coming from to get written to RCDB?

[sdobbs]

the trigger bits are set by the simulation of the calorimeter energy sums. the simulation takes most of its parameters (energy thresholds, channel masks, etc.) from the RCDB so that it can properly emulate the running conditions

[sdobbs]

so to clarify, if the relevant configuration files aren't loaded in the RCDB, then the can't be set in the simulation

[DraTeots]

Well yes, that is what I meant, probably those are configuration files. Are they saved for those runs? What are they? If they are not saved, are they still alive?

[nsjarvis]

I recalled the 4 run config files from tape. They are in
/work/halld/njarvis/runconfig_2017

30390 already has its config file loaded into RDCB, but in my analysis it showed the same problem as the other 3 runs, which do not have their files loaded into RCDB.

The tar file containing each config file also had many other files, eg current_run.log, info.txt, LMK If you need any of those as well.

[nsjarvis]

Hao's complaint here mentions a LIST file missing. I can't see that in the RunLog collection. Is that created when the config is loaded?

[sdobbs]

Well yes, that is what I meant, probably those are configuration files. Are they saved for those runs? What are they? If they are not saved, are they still alive?

Presumably the files being references are the main configuration files that Naomi pulled onto disk. Looking at the code, it's not entirely sure to me what it being looked for, since I don't understand how some of these RCDB keywords map into the actual DAQ files, but I've copied the relevant sections below, which I hope helps.

In general, all of the config files for each run were copied to tape in a tar file, so for the 2017-01 run period, you can look for /mss/halld/RunPeriod-2017-01/rawdata/RunRRRRRR/RunLogRRRRRR.tar - I think that's all of the files whose parameters would nominally get loaded into the RCDB.

auto  rtvsCnd  =  connection.GetCondition(runnumber, "rtvs");
  
  if( !rtvsCnd ) {
    cout<<"'rtvs' condition is not set for run " << runnumber << endl;
    return 2;
  }
  
  
  auto json = rtvsCnd->ToJsonDocument();               // The CODA rtvs is serialized as JSon dictionary.  
  string fileName(json["%(config)"].GetString());      // We need item with name '%(config)'

[DraTeots]

The files are identified by absolute-path -- run. So in this example string fileName(json["%(config)"].GetString()); is correct. But then one needs to read this file from DB, not from disk. Or, having a full file name one can guess where this file might be now (mss?) but that is about RCDB then.