100 likes | 237 Views
Post-Mortem Analysis of first Commissioning Week. Online Meeting 25 February 2008 Beat Jost/Cern-PH. Achievements. We DID actually read some fraction of many detectors About 60 Tell(UKL)1s… Surely a nice achievement… Congratulations!!!!!!!!!!!. Action List Legend.
E N D
Post-Mortem Analysis of first Commissioning Week Online Meeting 25 February 2008 Beat Jost/Cern-PH
Achievements • We DID actually read some fraction of many detectors • About 60 Tell(UKL)1s… • Surely a nice achievement… Congratulations!!!!!!!!!!! Online Meeting, 22 January 2008
Action List Legend Highest Priority (implementation <1 Week) • AS = Alba • CG = Clara • RJ = Richard • OC = Olivier • RS = Rainer • SK = Stefan Medium Priority (implementation <2 Weeks) Important (implementation <3 Weeks) Embellishment (implementation eventually) • EvH = Eric • DG = Domenico • NN = Niko • RSt = Radu • MF = Markus • BJ = Beat Online Meeting, 22 January 2008
Thing to do… • Controls • Reset should stop Tell1s from answering to TFC etc… (RS, SK) • Reset is not implemented in all menus (SD menus in particular) so one can’t always recover (CG, AS) • Ownership is not propagated always, in particular when starting a menu from the FSM panel (where one navigates to find the faulty guys). For example the farm menus are not owned by default. (CG,AS) • Trigger options (MEP packing, trigger source) interesting only when Active (AS, CG) • Dynamic allocation to be implemented everywhere (TFC, Farm, Monitoring, Storage) (RJ, CG, EvH, MF) • Clear selection of the content of the partition (detectors), available only when not_allocated (CG, AS) • Change of partition content (e.g. disabling TELL1) not available except when not_allocated (CG, AS) • When a transition doesn’t answer, the control is impossible. Is it possible to have a ‘Cancel’ command that would unconditionally stop the action, and allow e.g. a RESET (CG, AS) Online Meeting, 22 January 2008
Thing to do… • Control (cont’d) • Why is TRG in a different control panel than any other detector? Should be similar? (CG, AS) • Options for farm configuration (TAE, passed fraction) and storage configuration available only before Configure (CG, AS) • Trigger rate and dead time interesting only when Active / Running (AS, CG) • For that, maybe an overlaid panel in the main panel can allow these three configuration • Propagation of selection / exclusion not always working (CG) Online Meeting, 22 January 2008
Thing to do… • Farm Control • logViewer display twice most messages, at least when started with “–l 3”. (DG, NN) • Tasks should be properly terminated, and not killed by an external signal. This allows exit handler to be run properly (EvH) • When MEPRx dies, its control PVSS goes to OFFLINE but nobody goes to error -> run continues but all events are lost.(EvH) • Other DAQstuff… • Eventbuilder: • time-out for incomplete events + a PVSS panel to show which source shows which problem (NN) • an injector using (captured or written) MDF files to replay problematic sequences. (MF, NN) • TELLs: • configuration of the data-interface IP should come centrally (from the ConfDB?) and not be hardwired in individual recepies (host lookup of name-dx) (RS, SK) Online Meeting, 22 January 2008
Thing to do… • TFC • Move all detectors to dynamic allocation ASAP ! Should avoid the needs for manual changes. (RJ, CG) • MEP request scheme to be put operational (RJ, NN) • Implement throttling properly (RJ, BJ) • Use of ‘time alignment’ bit between L0DU and TFC to be revisited. We cannot mix normal events and TAE events in the same run: RICH incompatibility and farm configuration (RJ, OC, L0 Boys) • Occasional configuration error in ODIN. It is a timing problem in PVSS when checking that what was written to the board was actually what was requested. It happens usually after system reset. It is not serious but it is really annoying and should be solved (RJ) • Put in Event time (RJ) • The simplified TFC control panel to be added to the top ECS. (RJ) Online Meeting, 22 January 2008
Thing to do… • Storage • Run database... (RSt, MF, CG) • Clear report of the name of the written file in the main menu. (RSt) • The monitoring screen of Markus seems to double count. (MF) • Markus has fixed the writing problem, to be confirmed (MF) Online Meeting, 22 January 2008
Thing to do… • Monitoring • Central monitoring and basic event quality checking. (BJ, OC) • Nothing really tested so far, SD should provide tasks. (SDs) • Should devise a strategy for events that are mal-formed (bad bank structure, etc.) (OC, BJ, NN, MF) Online Meeting, 22 January 2008
Thing to do… • Operations • Every transition failure should have an error report associated. (AS, all) • AES screen to be installed. What about its partitioning (AS) • Distributed MbmMon, distributed logViewer... (??) • Domenico’s LogViewer: useful default configurations could be added as shortcuts (both on the commandline and as "cklicable windows) (DG, NN) • Get TDET properly working (CG, AS, all) Online Meeting, 22 January 2008