Monitoring and Data Quality

Data Quality should be checked for each muon sub-detector via OHP (Online Histogram Presenter).

Table of Content

Data Quality Applications and Tools

OHP - Online Histogram Presenter

The Online Histogram Presenter OHP is started by clicking on the button "OHP" on the DAQ Panel . Make sure you are using the muon configuration as specified here. Once the tool is open, you can
  • Navigate between the muon sub-detectors by clicking on the corresponding item in the left hand side column "Plugins", and ticking the box "XXXGlobalView".
  • Navigate between groups of histograms by selecting the corresponding tab
  • Enlarge a histogram by double clicking on it, this will open the histogram in a separate pop-up window. Note that the new window usually starts "minimized", ie appears only in the bottom menu bar of opened windows on your desk, click on it there to display it.
  • Zoom, change scale, etc. of a histogram via left and right mouse click operations, the OHP histograms are root objects.
Trouble Shooting
  • Histograms are missing: Check that a run is ongoing. Is the concerned sub-detector in the current partition ? Check the DAQ configuration (run control tree from the TDAQ IGUI), are the Gnam applications for the concerned sub-detector enabled and running ? If a Gnam application appears absent, ie has died, it can be restarted, ask the run control shifter to do so.

DQMF/D - Data Quality Monitoring Framework/Display

GNAM is an Online monitoring structure that allows monitoring of detector status and running conditions. It is implemented at the ROD level and it produces plots for online monitoring. The plots can be seen via the displays OHP and DQMD described here below. GNAm errors appears on the DAQ Message Reporting System MRS. The DQM Display (DQMD) allows shifters to browse through a subset of Gnam histograms, which are automatically tested and flagged for problems by the Data Quality Monitoring Framework (DQMF). From the DAQPanel, make sure the database file, Setup script and partition name are correct for the running partition, and then click on the "DQM Display" button on the "Main" tab on the right

MDT GnaMon

To use gnaMon, proceed as follows:

  • Start gnaMon from the Muons --> Monitoring/DQ menu via the MDT GnaMon item.
  • Add the current TDAQ partition as data source. You can do this either from the ganMon menu bar via File --> Add data Source --> Online Partition or clicking on the 'Data Source' button on the 'Data Sources' window. Type in the partition name and hit enter.
  • Verify HitsPerTube_ge_ADC has been chosen as histogram name. This information is displayed at the bottom of the GnaMon main panel. If you find a different setting, change it from the Tools --> Preferences menu.
  • Start auto-update for histograms clicking on the 'Auto-Update' button
  • Alternatively, you can request histograms to be refreshed/updated by Gnam pressing first the 'Publish histograms' button , then the Process button .
  • Switch to the 'Overview' panel which shows various chamber occupancy parameters, according to the selection in the drop down menu at the top of the panel, as color map. Double-clicking on a chamber opens the 'Details' panel with more information.

For further details, please also refer to the gnaMon online help which can be accessed from the gnaMon menu bar.

RPC DQ panel in DCS

What to Check at Start of Shift

What to Check during Data Taking: OHP



There are two tabs with a total of 24 histograms, each of which is important and must routinely be checked by eye during the shift, especially after stable beams is reached and HV is ramped to nominal.

  • Global Monitoring Overview
    • Occupancy Plots: One for Barrel, one for Endcap. Check for hot spots or holes. If you find one, cross-check it with the Whiteboard and, if it is not there, insert it and list the chamber in the shift summary.
    • TotalNumber_of_MDT_hits_per_evt_ADCCut:
    • Overall TDC vs. ADC Spectrum: The ADC vs TDC for all channels of the MDT. The peak of this distribution should occur at (100<ADC<160 and 400<TDC<800. And the distribution should tail off nicely from this peak region. Peaks outside of this region above the noise band (horizontal line around ADC=60) could signify gas problems in the MDTs.
    • MDTTDC_Summary_ADC: One summary plot of the TDC distribution for the BA, BC, EA, and EC. These histograms should be examined in the same way as the following tab's...

  • TDC Summary All histograms should be present and each should have a t0 in the range of 500-600 TDC counts and a tmax near 1300 counts. Look for any spikes in the distribution. If RPC triggers are off this will likely be noisy. If these are noisy and RPC triggers are on there may be a gas problem.
    If a histogram appears to be noisy or have an atypical t0 or tmax, report it in the shift summary.
    If one of the 16 histograms is empty or missing, cross-check it with the DQ shifter and inform the expert on call.


There are 4 histogram tabs. The data integrity tab should be always checked, the Hit maps especially after stable beams was reached and HV ramped to nominal.
  • Data Integrity . The first 4 histograms should be empty, the others should show a flat distribution. If not the case, submit a separate entry to elog with RPC+LV1+DAQ+Monitoring as affected system and specifying the run number. Please attach a screen shot.
  • Time . For proton collision runs, hit time distributions should be centered at zero and Gaussian like in shape.
  • TriggerHits . The top 2 plots are trigger hit maps, showing which RPC towers are active in the low-pt and high-pt trigger. You should identify any holes and compare them with known problems as listed on the muon whiteboard. If a tower is missing but not listed there, investigate in the RPC Rate and Busy panel if it was auto-killed and in DCS if there is no HV or similar.
    Report the missing tower in your shift summary, with the reason if you could identify it. If a larger region is missing instead of individual towers, or the number of un-understood towers is higher than 5, please call the RPC expert rather than waiting with reporting it for your shift summary. In the shown example, the 2 towers circled in red are missing, all others are present. The plot can serve as reference of expected coverage.
    SL output Hit Map shows hits sampled from the sector logic. Check for missing towers, report these in your shift summary. If towers are only missing in the SL plot but not the trigger hit ones, as in the shown example for trigger sector 27, there is no need to call the expert, even if multiple towers are affected.
  • Hits. These histograms are occupancy plots for the 6 individual RPC layers (BO stations confirm 0,1, BM station pivot 0,1 and confirm 0,1) as in the readout. Problematic towers are normally masked from both readout and trigger. Check this plots for holes, especially holes in the same spot for several layers, and for larger regions.



What to Check during Data Taking: DQMF



  • Click on the "MDT" button in the "DQMD Summary" window
  • Look at the "DQMD Details" window, and look for red flagged chambers in the display
  • Click on any chambers flagged red or yellow, select the "Histograms" tab for that chamber, and find out which histogram has been flagged red (i.e. has a red border)
  • Note in your Shift Summary elog the cause of each red or yellow flag, noting
    • The chamber name
    • The histogram name
    • The reason for the red flag (see the "Results" sub-tab under the histo itself) -- please include the numbers!
  • If many histos have red flags, put in a separate elog immediately, or call expert. (Attaching a screenshot to the elog can be helpful.)



Shift Online Data Quality Flags

Deciding Shift Online Data Quality Flags

Each run of more than 100000 events should be flagged by the muon desk shifter, to indicate whether the data is usable by each segment (MDTBA, MDTBC, MDTEA, MDTEC, RPCBA, RPCBC, TGCEA, TGCEC, CSCEA, CSCEC):

  • GREEN: At least 90% of the detector segment is performing well, has nominal operating conditions (LV, HV, Gas), and the data looks good
  • YELLOW: At least 90% of the detector segment has nominal operating conditions (LV, HV, Gas), but it looks like it might have bad calibrations, and could be fixed by reprocessing.
    In the case of muons, a yellow shift online flag should be put if the detector situation was unclear for reasons other than low statistics, typical examples are if the LV, HV etc. status in DCS was UNKNOWN due to a OPC server down or such.
  • BLACK: The segment was disabled or no data was taken from it
  • GREY: Data quality cannot be assessed (e.g. if there are not enough statistics)
  • RED: Bad data quality or no data is received from at least 10% of the detector, or detector was in STANDBY state (e.g. with HV not at nominal level)
The flag can be entered for each segment for each luminosity block separately. So a typical run might be flagged red at for some luminosity blocks for some segments if beam was being ramped, but then green later, after stable beams flag was set and the detector HV was ramped up to nominal.

Please enter a comment to explain your choice of flag, unless you are flagging green.

Entering Shift Online Data Quality Flags to the Database

To set the muon desk shifter data quality ('SHIFTONL') flags for recent runs,

  • Go to the "Last runs in the ATLAS partition" section of the MuonSpectrometerWhiteBoard. Check the run you want is listed, and for that run click on the boxes (any of the boxes will do) in the SHIFTONL column. This will open a new browser window with the DQ flagging interface.
  • Next you must set the colours you want, before then saving changes to the database. You can set a flag colour for e.g. MDTBA, by clicking on the column header titled 'MDTBA'; alternatively you can set the same flags for multiple segments by clicking on one of the higher-level headers, e.g. 'MDT' or 'Muons'. Clicking on such a column header brings you to a new page, where you must select the colour you want to flag. You should also enter a Luminosity Block range (or leave range boxes blank if you're flagging the whole run with the same colour), and you must enter a comment to explain the choice of flag (unless you're flagging green).
  • Next click 'Save Change' (but this button does not actually save the change into the database). You will be taken back to the page from which you can click on a header again, to enter a flag for a different detector segment or lumi block range.
  • Once you've selected flags for all the subsystems and luminosity block ranges in the run, click on the 'UPDATE DB' button. You will then be asked for a password, enter 'Muons', and then the new flags will be stored in the database.

Trouble Shooting
If the run you want to flag is not listed on the MuonSpectrometerWhiteBoard or clicking on the SHIFTONL boxes in the list there does not work for any reason, you can go directly to . Select Muons from the Subsystem drop-down box. Enter the run number you want. Choose DB instance 'COMP200 SHIFT ONLINE' (and leave tag as DEFAULT). Then click on "Display Info". A new window will appear, in which you should follow the instructions above.
Edit | Attach | Watch | Print version | History: r7 | r5 < r4 < r3 < r2 | Backlinks | Raw View | Raw edit | More topic actions...
Topic revision: r3 - 2011-04-04 - MarcellaCapua
    • Cern Search Icon Cern Search
    • TWiki Search Icon TWiki Search
    • Google Search Icon Google Search

    Sandbox All webs login

  • Edit
  • Attach
This site is powered by the TWiki collaboration platform Powered by PerlCopyright & 2008-2021 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
or Ideas, requests, problems regarding TWiki? use Discourse or Send feedback