DST event number range overlaps causes query mis-matches

[c-dilks]

For some DST files, we find that evnumMax of DST file N is larger than evnumMin of DST file N+1 (more correctly, of N+5, since these are 5-files). Thus the boundary between files N and N+1 becomes blurred: events in the "overlap" of DST files will be associated to two DST files, since QADB finds the DST file by checking event number ranges. In practice:

• QADB queries by (runnum,evnum) will return the first matching DST file that contains the event
• Accumulated charge calculation may be slightly overestimated for runs that have such files

To determine which runs and DSTs have this issue, use util/syncCheck.groovy:

source env.sh
run-groovy util/syncCheck.groovy  # prints all the overlap info
run-groovy util/syncCheck.groovy | grep 'SYNC ERROR' | cut -d= -f2 | sort -n | uniq -c

The third command prints two columns, shown in the list below: the number of problematic DST files, together with the run number.

In some cases, only one run has an issue; sometimes the difference evnumMax - evnumMin in a DST file is very large, causing that DST file to be associated to all events of a run. In other cases, the entire run has overlapping DST files.

It is not clear how to fix this issue from the QADB side. A possible upstream solution is to ensure during cooking that the events do not get "mixed" between DST files.

      1 5126
    237 5128
    101 5129
    238 5130
      1 5163
      1 5194
      1 5198
      1 5203
      1 5233
      1 5261
      1 5342
      1 5357
      1 5358
      2 5362
      1 5366
      1 5368
      1 5431
      1 5507
      1 5520
    238 5525
    240 5526
    346 5527
    336 5528
     18 5564
      2 5565
     34 5566
     67 5567
      1 5569
      1 5717
      1 5749
     10 5827
      1 5848
      1 5904
      1 6298
      1 6333
      2 6378
      1 6379
      1 6381
      1 6433
      1 6443
      1 6472
      1 6479
      1 6481
      1 6501
      1 6513
      1 6547
      1 6565
    261 6631
    126 6632
      1 6647
     25 6656
      1 6676
      1 6685
      1 6710
      1 6716
      1 6731
    363 6775
    326 6776
    391 6777
    322 6778
      1 11127
      1 11210
      1 11290
      1 11340
      1 11403
      1 11417
      1 11480
     22 11504
      1 11517
      1 11557
      1 11568
      1 11571

[c-dilks]

Note: util/printSummary.cpp marks some DST files as Golden, but they do not appear in text/listOfGoldenFiles.txt. This is likely related to this issue. The runs and filenums are:

5163 125
11557 590

Both of these runs appear in the list above.

[c-dilks]

Newer QAs have been using time bins, which do not have this issue. We will not be going back and fixing old Pass 1 QAs for run groups A, B, or K, since their Pass 2 QAs will be done with time bins.