Minutes of Trigger DB/L2 Software Working Group Meetings

Specifications Generated at DB Meetings

Mar 5, 2002

Attendees:

        Jonathan, Alexei, Donatella, Peter, Stephen (phone), Tom (phone), Peter Wittich, Masa, Luciano

GUI News (Donatella):

Feb 28, 2002

Attendees:

        Bill Badgett, Peter, Donatella, Peter, Kirsten

Special Meeting to discuss New DB change implementation:

Feb 19, 2002

Attendees:

       

GUI News (Donatella):

Feb 12, 2002

Attendees:

        Jonathan, Alexei, Donatella, Peter, Stephen (phone), Tom (phone), Peter Wittich, Masa, Luciano

GUI News (Donatella):

Status of New L2 Options (Alexei):

Plans for L2 Tagset and Global Parameters

Feb 5, 2002

Attendees:

        Jonathan, Donatella, Peter, Stephen (phone), Tom (phone), Peter Wittich, Masa, Karen

GUI News:

New L2 Electron/Photon/RECES options:

Status of new L2 code (update from last week):

L2 Tagset:

Jan 29, 2002

Attendees:

        Alexei, Jonathan, Donatella, Peter, Stephen (phone), Tom (phone), Peter Wittich, Masa, Karen

News:

Status of GUI - Donatella

Electron and Photon Options and Code - Alexei + Masa

Rate Limit and Prescale

Plans for new L2 code

AOB

Jan 22, 2002

Attendees:

        Alexei, Jonathan, Peter, Stephen (phone), Tom (phone), Peter Wittich, Masa, Karen, Nathan

Electron and Photon Options and Code

L1 Track Trigger

L2 Isolated Track Trigger Spec

ClusterSum Spec

Dec 11, 2001

Attendees:

        Jonathan, Donatella, Peter, Kirsten, Stephen (phone), Peter Wittich

News:

Status of GUI Merging:

Status of New L1 Calorimeter Options:

Dec 4, 2001

Attendees:

        Jonathan, Donatella, Alexei, Peter, Heather, Nathan

News:

Status of GUI Merging:

Status of New L1 Calorimeter Options:

Plans for L1 Track Trigger Options/Code and new cuts and parameters:

Nov 27, 2001

Attendees:

        Jonathan, Donatella, Alexei, Kirsten, Peter, Heather, Nathan, Stephen

News:

Discussion of how to handle XFT pt values

Nov 20, 2001

Attendees:

        Jonathan, Donatella, Alexei, Kirsten, Peter

News:

Discussion of Trigger Reports

  • Donatella's summary in comma separated format (input to spreadsheet such as EXCEL or Gnumeric) will be written to directory parallel to Alexeis.  Write this for all Validated tables.
  • Alexei's reports
  • Randy's DB browser
  • Run Summary page
  • Page with list of Physics Tables that were/will be used for Physics data taking (not test tables or ones that were only used once).  This will have symbolic links to copies in Alexei's and Donatella's lists
  • Need to develop improved "standard" views for Randy's browser

    • Nov 13, 2001

    Attendees:

            Jonathan, Donatella, Alexei, Heather, Peter
     

    News:

    Changes to L1 and L2 Options

    Nov 6, 2001

    Attendees:

            Jonathan, Donatella, Kirsten, Alexei, Dan (speakerphone), Peter
     

    News:

    Plans for GUI Merging:

    Agenda for Next Week:

    Oct 30, 2001

    Attendees:

            Jonathan, Donatella, Kirsten, Stephen, Heather, Peter W, Peter W,
            Rick St Denis, Randy Herber, Jack Cranshaw

    News:

    Problem of Mixed case information in database:

    Specifcation of L2 Electron, Photon and Two-Track Code interfaces:


     

    Oct 23, 2001

    Attendees:

    News:

    Discussion of Global Parameters:

    Date: Mon, 22 Oct 2001 17:19:15 -0500
    From: Peter Wilson <pjw@fnal.gov>
    To: Trigger Database Working Group <amidei@umich.edu>,
         Frank Chlebana <chlebana@fnal.gov>, demers@fnal.gov, hray@fnal.gov,
         jdl@fnal.gov, jsm@fnal.gov, kirsten@fnal.gov, ksmcf@fnal.gov,
         luciano@fnal.gov, meyera@fnal.gov, myron@umich.edu, patrick@fnal.gov,
         smaria@fnal.gov, torretta@fnal.gov, varganov@fnal.gov, wittich@fnal.gov
    Subject: Global parameters and a few other things to add to the Trigger
        database

    The general idea is to add a new table to the Trigger database that would
    contain a list of global parameters.  The table that spaecifies the
    Physics table would have the version of the set of global parameters added
    to it.  That way if we add a new global parameter we can be backwards
    compatible.

    Here is the list of global parameters that Jonathan and I came up with.
    This will be a topic of discussion in tomorrows trigger database meeting.

    Peter

    1. L2 Code CVS Tag - this would be the tag for the code that L2 was
    compiled against.  In the GUI you would be able to select "Current" or
    specify a specific version for testing.

    2. L1 Calorimeter/DIRAC Tower Granularity - currrently selected for each
    trigger.  This would be a single value for HAD, EM, Plug and Central.

    3. L1 Calorimeter/DIRAC HAD/EM granularity (aka HADEMsel) - again
    this is currently selected per trigger.  There would be a plug and a
    Central version (two parameters).

    4. Set number for XFT Pt bins - this may be tied to the hardware DB.

    5. Set number for SVT roads - similar to XFT bins.  It is not clear that
    this is needed for the trigger table.

    6. Firmware versions for Muon trigger cards:
            a) Match cards
            b) PreMatch cards
            c) Muon Trigger Summary

    7. Firmware version(s) for XTRP firmware?

    8. CSX Trigger using MeanTimers or Single Counters

    9. Association of Muon delta t with pt threshold.  Muon Delta T cuts?

    These we think should go into or stay with the individual triggers:

    1. DCAS seed and shoulder thresholds

    Other items that should be added or start being used but not global
    paramters:

    1. For predicted trigger cross-section: add contant rate and non-linear
    terms (1 parameter goes to 3)

    2. Start filling information on valid rages for parameters in the datbase.
    What needs to be done to implement this?  Who can do it?

    3. Implement rate limit option at L2?

    4. Can the L1 prescale specific option be extended to a L1 Dynamic
    prescale option with multiple parameters?  Does this require a schema
    change?
     

    Items that should go into the other databases that might not be there:

    1. Calibration constants for RECES thresholds in calibration database.
    This is similar to ADMEM getting calibration constants used for Trigger
    Tower energies.
     
     
     
     

    Oct 8, 2001

    Attendees: Jonathan, Donatella, Kirsten, Peter, with Dan and
    Stephen on speaker-phone

    Status of Error handling: Alexei could not report as he was retrieving his
    car.  We will discuss this further next week.

    Donatella reported on what she see's as her current priorities:
    1. Fixing DB connection issue with Alexei's DB - with Alexei
    2. Merging calls to Alexei's L1, L2 pages directly from her page
    3. Pop up window of validation errors from L1/L2 trigger validation when
    making tables
    4. Check for duplicated L1, L2 or L3 triggers with different version
    numbers before committing Physics table
    5. Fixing ordering by version number in GUIs

    We then discussed the overall priorities for DB work by Nov 17 (end of
    shutdown):
    1. Make GUI/table making ready for less expert people to make tables
      o Finish GUI merging
      o Improved error reporting for L1/L2
      o Written procedure on web for making tables
            - Need to clarify Oxford operations: need better contact such as a
            single mobile phone number to call or a clear shift rotation.
      o Moving of TCL files to code browser needs to be automated
      o Need to retire old Options, triggers, paths, datasets and tables

    2. L3 Filters moved to new Filter Package in offline.  Under librarian(s)
    in L3 group.  This was later discussed in the TDWG meeting on Thursday.

    3. L2 code/executable versioning: Stephen said he and Peter Wittich would
    come up with a plan.  This will be presented in 1-2 weeks.

    4. Switch from TFRD to TL2D for the L3 prereq module.
    After experience from the past few weeks it has become clear that we do
    not want to switch back and forth between TL2D and TFRD for cutting/tagging
    runs.

    The problem:
    We definitely want to use TL2D for the rereq module when we
    are cutting at L2.  Otherwise there are volunteers at L3 that increase
    rates.  They also make it difficult to understand the rates from the
    monitoring programs.  However, with the current scheme of switching from
    cutting to tagging at L2 there is a problem with using TL2D in tagging
    runs.  The effect of doing tagging at L2 and using TL2D for the prereq at
    L3 is to make the L2 decision in L3 using the TL2D bits.  This defeats
    the purpose of L2 Tagging.  One option is to switch to TFRD for the prereq
    module if we are tagging at L2.  This is was the proposal put out by
    Kirsten in early September including several options for implemenation.
    The problem with using TFRD is that we get rates for the L3 triggers
    that are very different in L2 tagging than L2 cutting so it is
    difficult to validate the table.

    The proposal:
    1. Switch to using TL2D for any trigger table that has L2 triggers
    (essentially all tables).  L3 would only look at the TL2D bits for
    a given path.  L3 will assume that L2 has already required the correct L1
    bit for that path.  Making sure that that is done correctly will be left
    to monitoring (see L3 Filter for TLD errors in item 6 below).

    2. Only special L1/L3 only tables would use TFRD.  This would be done by
    using a manual switch by Kirsten/Sarah as to which TCL template to use.

    3. Add a L2 monitoring path to the physics tables:
            Level 1 an OR of all the L1 triggers in the table
            Level 2 an auto-accept
            Level 3 run all reconstruction but no filter

            For L2 cutting, L2 would have prescales to make the
            rate of the L2 Monitoring path be on the order of 1% of the total
            L2 output rate.  There would be two groups of triggers with
            separate prescales: track triggers and all others.   The track
            triggers would get a larger prescale.

            For L2 Tagging, the L2 prescales would be set to 1.

            Tagging vs cutting would be set by RC.  If in tagging mode RC
            would send down prescales of 1 instead of the prescales in the
            TriggerDB.

    These next items are not directly part of the switch to TL2D but are
    related to the L2 monitoring path in their needs.

    4. Add a L3 monitoring path to the physics tables:
            Level 1 don't care
            Level 2 an OR of all the L2 triggers
            Level 3 run all reconstruction but no filter

            L3 would have prescales to make the
            rate of the L3 monitoring path be on the order of 1% of the total
            L3 output rate.

    5. Add a L2 Error trigger path to the physics tables
            Level 1 an OR of all the L1 triggers in the table
            Level 2 would be controlled by the L2 Error Handler
            Level 3 run all reconstruction but no filter

            The L2 error handler would set this trigger for events which have
            problems that corrupt this events data but not future events.
            There would be a mechanism to have an escalating prescale
            (e.g. take first 10 events, then take 1/10 for next 100, then
            1/100 for next 1000 etc) for each error type.  Would include error
            flag in data (TL2D or error bank?).

    6. Add a L3 Path (and filter) for TL2D errors to physics table
            Level 1 don't care
            Level 2 an OR of all the L2 triggers
            Level 3 do comparision of TL2D with TFRD, TL1D, Clusters, XFLD,
            SVTD to make sure that all data got into processors corectly.
            Accept events with corrupt TL2D info

            As with the Error handler trigger at L2, there should be a
            mechanism for an escalating prescale so that the bandwidth is not
            saturated by these errors.  Of course choking the system would be
            one way of alarming the shift crew that there is a problem with the
            trigger.

    7. For items 3, 4, 5, and 6 would like to have a monitoring stream out of
            Level 3.  This should be small for good running with L2 and L3
            cutting.  It would get large if there were major problems with L2
            or if we in L2 tagging run.

    Oct 1, 2001

    Attending: Alexei, Jonathan, Donatella, Peter Wilson, Dan (speakerphone)

    Reports of on going work:

    Alexei is working on several L2 issues: Error Handler and L2 SVT Trigger.
    He expected to finish these on Monday.  His next job is to work on the
    database connection to the L1/L2 GUI.  We have several cases of switching
    back and forth between Alexei's L1/L2 GUI and Donatella's
    Paths/Datsets/Tables GUIs causing problems with the database.  Alexei will
    try to make a connection to the database handles that Donatellas GUI opens
    rather than opening a new one.  This is critical to proceeding with GUI
    merging.

    Donatella has made several improvements to the Physics Table GUI.  It now
    looks for duplicated paths in the Physics table and issues a warning
    before committing the information to the database.  The user can then
    continue like you would want to do for paths appearing in Express as well
    as another dataset or return to the Table editor and fix the problem.  The
    check issues a warning wether the paths have the same version number or
    not.  She is looking into doing the same kind of check to require that all
    copies a a given trigger (at L1, L2 or L3) have the same version number.
    Next on Donatellas list is merging the GUIs.  She will start by calling
    Alexei's L1 and L2 panels directly from her panel instead of going through
    Alexei's intermediate panel.
     

    We had a brief discussion on what issues to tackle next:
    1. GUI merging.  We will need to resolve several issues to make
    significant progress:
            - Need Alexei to fix DB connection
            - Need to decide on global parameters

    2. Error reporting for table validation (L1/L2)

    3. Planning for what we want to have implemented before Nov 17 (end of
    shutdown)
            - TL2D for l3 prerequiisties
            - GUI Merging complete
            - L2 Code management
            - Global parameters

    Last updated by Peter Wilson pjw@fnal.gov on 19 Nov 2001

    Created by Peter Wilson pjw@fnal.gov on 6 Nov 2001