1. Online Control Program: a summary of recent discussions
R. Fantechi
8/2/2012

2. Kick-off meeting Held on Jan 16th, via EVO People from NA62 and EN/ICE
Following a first meeting (Riccardo, Marco, Valeri, Ferdi) with EN/ICE (Manuel Gonzalez, Fernando Varela)
EN/ICE willing to provide expertise, but dedicated manpower from NA62 needed
A first implementation with maybe reduced functionality for July
List of requirements extracted from note NA added informations
Nicholas Lurkin from Louvain will work on the CP.
Needs to become expert, i.e. through standard EN/ICE courses
Description of the NA48 Control Program functionality demande to a discussion between EN/ICE and myself
The scope was extended to let myself learn what LHCB is doing

3. The NA48 control architecture - 1
Hardware
VME single board computers + Ethernet
Separate data and control path
One case for a PC (farm control) as control equipment
Finite state machine
Run states handled by the so called “harness”, a process in each control machine
Receives state changes from the Control Program
Executes user defined programs for any state change
Maintains the integrity of the states
Communication
Message passing through ControlHost (developed by Maslennikov and Gurin)
Similar to DIM, used in LHCB and in PVSS control systems

4. The NA48 control architecture - 2
Control program
Implemented with a TCL/TK scripts
Both state machine and GUI
Run database implemented as an ASCII file, later transformed into HTML
Interface to the configuration database
Simple definition of states, forming a cycle
Idle, Init, Ready, Running, EndRun, Terminate
Configuration database
Implemented with Objectivity (overkill…)
Configuration files from the DB read at initialization time through a GUI started by the CP
The files for the selected configuration sent to a common NFS mounted area where they were read by the various SBCs during the init phase
Messaging
Use of EMU to centralize message handling/display
This list defines the basic requirements of NA62 too

5. The LHCB control suite - 1
Components
A run control machine running PVSS and SMI++
A number (~40) of control PCs with PVSS and SMI++ to control the Event Filter Farm
The large number of PCs allows to slice the farm in sub-farms which could be allocated to different sources if the readout is partitioned
Similar PCs for each detector
Sometimes one PC for multiple detectors
They have LHCB specific PVSS components to interface the CCPC for TELL1 configuration (via a DIM server on CCPC)
Different type of hardware (i.e. VME/Cream) could be handled with the same generic mechanism, where the server is run on a PC linked to the VME system
Similar PCs for storage and trigger control
All these computers are linked via PVSS distribution mechanism. Error/info messages handling is embedded in PVSS

6. The LHCB control suite - 2
Configuration
Configuration data stored on Oracle DB
A PVSS interface called “Hardware tool” to handle these data
GUI to define all the boards/registers
One to one mapping of TELxx registers to PVSS data points
The specific PVSS components for the TELxx handle automatically the retrieval of data at init time
A lot of optimization done to speed up operations…
Run control
The same kind of states as in NA48
Integrated environment
It also allows for partitioning of the readout with concurrent operations of different sub-detectors in stand-alone mode

7. Control program for NA62 LHCB stuff could be reused
It needs customization, I am not able to quantify how much
Needs a detailed analysis with EN/ICE
Possible scenario
1 run control machine
1 farm control machine (could include storage)
Need to partition the farm?
1 or few PC for the detectors, running multiple instances of PVSS projects
With interface to TEL62, Cream control, GTK control?
The Oracle database

8. An alternative?
Clone the NA48 model implementing it with new technologies
Use of DIM
State machine handling to be written
Adapt harness scheme?
Which kind of GUI?
Handling configurations
Not easy to evaluate pro/cons
EN/ICE expert consultancy will probably drop
PVSS licensing will be simpler (only DCS)
More effort on our side?
Code maintenance problems
However keep in mind that a LHCB-like solution will imply some more strict maintenance of the operating system to follow upgrades in PVSS, which could also affect LHCB specific components