Transcript MonitoringCaloMtg21092011

Data Online Monitoring: Last News
B. Viaud
(LAL-In2p3)
Calorimeter Meeting, Sept. 21st 2011
1
Overview
• Data manager presenter pages;
• Pi0 online monitoring;
• The other things on the way.
Calorimeter Meeting, Sept. 21st 2011
2
Data Manager Shifter Presenter Page
• Reminder: we want to help the DM by reducing the number of items
she/he has to check. Also reduces the probability that something
missed by an overwhelmed DM.
• We studied the alarm pattern all over 2011 data to determine how to
group the many monitored quantities into only a few categories (see
my talk at the calo meeting 3 weeks ago).
• We modified accordingly the Shifter pages in the presenter.
Calorimeter Meeting, Sept. 21st 2011
3
DM Shifter Page
Calorimeter Meeting, Sept. 21st 2011
(XCAL)
4
DM Shifter Page
(XCAL)
Replaced with
Calorimeter Meeting, Sept. 21st 2011
5
DM Shifter Page
Calorimeter Meeting, Sept. 21st 2011
(PRS)
6
DM Shifter Page
(PRS)
Replaced with
Calorimeter Meeting, Sept. 21st 2011
7
DM Shifter Page (SPD)
Calorimeter Meeting, Sept. 21st 2011
8
DM Shifter Page (SPD)
Replaced with
Calorimeter Meeting, Sept. 21st 2011
9
0 Online Monitoring
• A first version written 5 months ago with "simple" trending plots (like
those produced for all the monitors)
• Information stored directly in root histograms;
• Somewhat slow and heavy, especially if one wants to check the trending
over more than a few hours;
• More relevant to use the new trending tool developed recently by O.
Callot (LHCb-INT-2011-003)
• Information stored in a ".trend" file. Can be decoded very fast, makes it
easy to build trending plots over very long periods ( ~1 second in the
presenter for a trend over month);
• Adapted the 0 trending algorithm to use this tool
• For the moment: trending of m(0) in Inner, middle, Outer, Whole;
• Other quantities, or zones (ex: those used for the fill-to-fill calibration)
can easily be added. We can also monitor ;
• The algorithm merges a user-defined number of savesets before a fit is
performed: allows to have that enough statistics;
• Still a few loose ends: hopefully installed at the pit within a few days.
Calorimeter Meeting, Sept. 21st 2011
10
Example of sensitivity (fit distributions merged over 5 Runs )
Sept. 12th 2011
-- Whole
-- Inner
-- Middle
-- Outer
Sept. 20th 2011
Sept. 12th 2011
Calorimeter Meeting, Sept. 21st 2011
11
Also on the way…
• Re-do the alarm survey presented last time with different alarm
threshold to (if possible) further reduce the number of problems to
be dealt with by the DM.
• Polish the scripts I wrote for that, so that they can be used by the
Piquets: provide once a day a summary of the alarms that triggered
during the last 24h.
• Add a monitor (and the corresponding alarm) for the crate relative
phase timing (cf. Frederic’s talk).
Calorimeter Meeting, Sept. 21st 2011
12