%CERTIFY%

InDetRTT Results

Introduction

This page is used for recording common issues or observations as a result of daily monitoring of the InDetRTT results. Please sign each item with your initials and, if relevant, date when it first appeared.

Monitoring

Daily monitoring of the InDetRTT results will be performed by the following ID Validation "shifters":-

  • 10-17 Sep 2007: Tim Adye
  • 17-24 Sep: Lauren Tompkins
  • 24 Sep-1 Oct: Bruce A. Schumm
  • 1-8 Oct: Tim Adye
  • 8-15 Oct: Lauren Tompkins
  • 15-22 Oct: Bruce A. Schumm
  • 22-29 Oct: Tim Adye
  • 29 Oct-5 Nov: Lauren Tompkins
  • 5-12 Nov: Bruce A. Schumm
  • 12-19 Nov: Tim Adye
  • 19-26 Nov: Lauren Tompkins
  • 26 Nov-3 Dec: Tim Adye
  • 3-10 Dec: Bruce A. Schumm
  • 10-17 Dec: Lauren Tompkins
  • 17-21 Dec: Bruce A. Schumm
  • 7-14 Jan 2008: Tim Adye
  • 14-21 Jan: Lauren Tompkins
  • 21-28 Jan: Bruce A. Schumm
  • 28 Jan-4 Feb: Lauren Tompkins
  • 4-11 Feb: Tim Adye
  • 11-18 Feb: Bruce A. Schumm
  • 18-25 Feb: Lauren Tompkins
  • 25 Feb-3 Mar: Tim Adye
  • 3-10 Mar: Bruce A. Schumm
  • 10-17 Mar: Tim Adye
  • 17-24 Mar: Stephen Haywood
  • 24-31 Mar: Bruce A. Schumm
  • 31 Mar-7 Apr: Lauren Tompkins
  • 7-14 Apr: Lauren Tompkins

Here is a suggestion of some things to check (TJA):

  1. Check for missing jobs. Could be temporary RTT problem.
  2. Check error logs for new messages.
  3. Compare a few of the statistics numbers (eg. number or efficiency of "Tracks" collection).
  4. Scan summary page for each new job listed on Seth's InDetRTT Plot Index (or, failing that, the InDetRTT results from RTT runs for the past week). Look at the top-level histograms to check that they looked OK by eye. I then looked in detail at a few representative ones, eg. Z->mumu efficiencies, fakes, pulls.
  5. Not yet much from the dCube results, because so many of the reference histograms are empty. In many cases, the plots seemed suspiciously identical to the reference. Can we compare against other releases to see if this is really true? This needs a closer look.

Please modify this list according to your experience.

Tools

idrttlogs

Usage:

~adye/bin/idrttlogs RELEASE [BRANCH]

Eg.

idrttlogs 13.0.20
idrttlogs rel_5 bugfix

Parses the logfiles from CERN AFS and creates InDetRecStatistics-RELEASE-BRANCH-DATASET.stat, .err, and .log for each of the six datasets. The .stat file contains the InDetRecStatistics summary report. The .err file lists all lines with WARN, ERROR, or FATAL (and some variants thereof), as well as any stack dump (run through c++filt). The .log file is a symlink to the original Athena job logfile (this will disappear after ~1 week). A symlink to the RTT results directory is also created (useful if idrttlogs is run in a CERN web-accessible directory).

IDRS_table_parser.py

Author: Seth Zenz

NOTE from Seth: This should work, but may become out of date from time to time. If it is having trouble, check the subdirectories under the click here for log file comparisons link on the main InDetRTT link page. Those are generally what I check when I update some aspect of the parsing functions.

Usage:

/afs/cern.ch/user/s/sethzenz/scripts/IDRS_table_parser.py [log file] [ref log file] [optional output file]

Eg.

/IDRS_table_parser.py InDetRecStatistics-rel_3-pcache-AthenaZmumu.log ../../13.0.30/InDetRecStatistics-13.0.30--AthenaZmumu.log

Parses the InDetRecStatistics log files and highlights in red (astericked) thoses numbers which differ from the reference log file by 20% or more.

Reporting

Report suspected problems by submitting a Savannah Inner Detector bug report (if that isn't possible, send the report directly to the Tracking and ID Release Coordinators, Markus Elsing and Andi Wildauer). A summary of outstanding problems, if any, should be given at the ID Software Weekly meeting (or posted in reply to the meeting announcement).

Weekly report of status (even if "all OK") to be added to the minutes of the Software Validation Reports virtual meeting. This should be sent by Monday noon (or Tuesday morning at the latest), Geneva time. These will be incorporated into the weekly Software Validation Report.

CC the other shifters in any Savannah reports, and also send them a copy of the Software Validation Report.

Discussion

Current areas of interest

  1. val and bugfix nightlies are not running at present. (TJA, 1 Oct 2007)
  2. pcache no longer need be checked (13.0.30.3 is expected to be the last production cache). (TJA, 14 Nov 2007)

Outstanding Issues

  1. Very few entries in hit residual plots for pcache "default tracking". This is due to track slimming on the default tracking output (for smaller ESD). This is fixed in 13.0.30 and dev (track slimming disabled in InDetRTT). (TJA, 1 and 29 Oct 2007)
  2. pcache misaligned 10GeV muons dynamic page is missing (so no link from InDetRTT Plot Index), but otherwise OK. (TJA, 22 Oct 2007)
  3. Improvement in the default tracking's fake rates in the dev nightlies WRT 13.0.30 and pcache nightlies. xKalman and iPatRec don't show any change. Hopefully this is expected and a Good Thing, but not heard back from Andi. (TJA, 29 Oct 2007)
  4. Warnings in dev nightlies, reportedly under investigation by Ed Moyse:
DetectorStore WARNING retrieve(default): No valid proxy for default object
ToolSvc.MuonEventCnvTool FATAL Could not get MuonGeoModel DetectorDescription manager
ToolSvc ERROR Error initializing tool 'ToolSvc.MuonEventCnvTool'
SegmentCollectionCnv WARNING Could not get MuonEventCnvTool
dissapeared for rel 3, back for rel 4
  1. Warnings in dev nightlies (Z->mumu) like
ToolSvc.InDetUpdator WARNING -U- track theta = -10.0881 -> this is too far from defined range, stop update.
ToolSvc.InDetUpdator WARNING calculateFS_5D(): bad angles in filtered state!
This is a known problem in the Kalman fit with updates becoming unstable. Wolfgang is trying to reduce this instability.

  1. Warnings in dev nightlies (all streams): [BAS 12/06/07]
Py:ConfigurableDb   ERROR Problem with [AtlasAnalysis_merged_confDb] content!
Py:ConfigurableDb   ERROR No module named ConfigurableDb
Py:ConfigurableDb   ERROR   ==> culprit is package [EventViewInserters] !
Py:ConfigurableDb   ERROR Problem with [AtlasAnalysis_merged_confDb] content!
Py:ConfigurableDb   ERROR No module named ConfigurableDb
Py:ConfigurableDb   ERROR   ==> culprit is package [EventViewDumpers] !
Py:ConfigurableDb   ERROR Problem with [AtlasAnalysis_merged_confDb] content!
Py:ConfigurableDb   ERROR No module named ConfigurableDb
Py:ConfigurableDb   ERROR   ==> culprit is package [EventViewTransformation] !
Py:ConfigurableDb   ERROR Problem with [AtlasAnalysis_merged_confDb] content!
Py:ConfigurableDb   ERROR No module named ConfigurableDb
Py:ConfigurableDb   ERROR   ==> culprit is package [EventViewBuilderAlgs] !
Py:ConfigurableDb   ERROR Problem with [AtlasAnalysis_merged_confDb] content!
Py:ConfigurableDb   ERROR No module named ConfigurableDb
Py:ConfigurableDb   ERROR   ==> culprit is package [EventViewCopiers] !
Py:ConfigurableDb   ERROR Problem with [AtlasAnalysis_merged_confDb] content!
Py:ConfigurableDb   ERROR No module named ConfigurableDb
Py:ConfigurableDb   ERROR   ==> culprit is package [EventViewCombiners] !
Py:ConfigurableDb   ERROR Problem with [AtlasAnalysis_merged_confDb] content!
Py:ConfigurableDb   ERROR No module named ConfigurableDb
Py:ConfigurableDb   ERROR   ==> culprit is package [EventViewUserData] !
Py:ConfigurableDb   ERROR Problem with [AtlasAnalysis_merged_confDb] content!
Py:ConfigurableDb   ERROR No module named ConfigurableDb
Py:ConfigurableDb   ERROR   ==> culprit is package [EventViewTrigger] !
and also
RelationalDatabase  Warning Release number backward compatibility - NO SCHEMA EVOLUTION REQUIRED: 
database with OLDER release number 2.2.0 will be opened using CURRENT client release number 2.2.2
and finally, in the Zmumu streams
ProtoTrackStateOnSurface >>> WARNING possibly bad use of ::checkinForwardPar!
and
ToolSvc.InDetBKS                                  WARNING extrapolation during backward filtering failed, 
reject track (activate DEBUG for more info)
This latter message has appeared in the past, eliciting the following somewhat ambiguous response from Markus Elsing (?):
this happens from time to time. It should not appear just like this...
cheers, markus
[end BAS 12/06/07]

False Positives

Resources

Talks from October 2007 ID Week Validation Session

  1. Tim Adye's Introduction.
  2. Lauren Tompkins' talk on The InDetRecStat Package.
  3. Seth Zenz's talk on Inner Detector RTT.
  4. Bruce Schumm skirts the issue of Back Tracking Validation.


Major updates:
-- TimAdye - 07 Sep 2007 -- BruceSchumm - 03 Oct 2007

%RESPONSIBLE% StephenHaywood
%REVIEW% Never reviewed

Edit | Attach | Watch | Print version | History: r1 | Backlinks | Raw View | WYSIWYG | More topic actions
Topic revision: r1 - 2008-04-02 - StephenHaywood
 
    • Cern Search Icon Cern Search
    • TWiki Search Icon TWiki Search
    • Google Search Icon Google Search

    Atlas All webs login

This site is powered by the TWiki collaboration platform Powered by PerlCopyright & 2008-2020 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