This website is no longer maintained. Its content may be obsolete. Please visit http://home.cern for current CERN information.
Minutes of the PM data format meeting. March 2006
Present: Mike, Greg, Nikolai, Dmitriy, Adriaan
Discussion:
Nikolai presented a plan to extend the present PM client API with a structure
that codes single data items and arrays with their description as a CMW object.
The binary dump with binX will be kept for backward compatibility. The advantages
are that conversion to SDDS of the new structure can be done by a generic
converter on the server and there is no need for the binX.
Everyone agreed that the next PM clients (RADMON and BI) should use the new structure and we should ask the existing clients (QPS and PC) if and when they can adopt the new data structure.
Conclusions:
1. Discuss the new PM data structure with the clients and check if they agree.
Meeting could be held during the first week of April.
2. Implement and test (Nikolai's estimate: 1 week + 1 week, after the agreement)
4th October 2005
Present:
"Boris Khomenko" <Boris.Khomenko@cern.ch>; "Kris Kostro" <Krzysztof.Kostro@cern.ch>; "Grzegorz Kruk" <Grzegorz.Kruk@cern.ch>; "Herve Milcent" <Herve.Milcent@cern.ch>; "Stephen Page" <Stephen.Page@cern.ch>; "Adriaan Rijllart" <Adriaan.Rijllart@cern.ch>; "Rudiger Schmidt" <Rudiger.Schmidt@cern.ch>; "Nikolai Trofimov" <Nikol.Trofimov@cern.ch>; "Markus Zerlauth" <Markus.Zerlauth@cern.ch>; "Verena Kain" <Verena.Kain@cern.ch>;
Alastair Bland reports that server CS-CCR-PM1 has already been setup and is available for use.
Nikolai explained that the V0 PM server was deployed and ready for testing. More details at PM Data Collection and Storage
The middleware threading problem had been examined with Steve and the problem fixed, at least in a simplified configuration. Compatibility with FESA also seem to be OK.
Nikolai will move his server to the new machine [since done].
PIC
No explicit trigger in the PIC but continuous buffering. Also no direct CMW interface, therefore PM system will go to the logging system for the moment to get the data as required.
Herve Milcent explained that the QPS system would attempt to interface to Nikolai's API after ICALEPCS
Greg Kruk gave a white board break down of the proposed directory structure for post mortem data - both raw and processed. He explained the proposed functionality of the PM converter/manager.
One issue that requires clarification is the notion of PM event. One suggestion was that the PM data could be store without reference to a PM ID and that a logical view of the stored data could provide the required convenience of associating given data with an event. (use of symbolic links etc.
Greg will implement a prototype structure and document said. This will include the approach to sequencing which will be dealt with.
Write to file and store with the rest. API to be provided by Alarm team. Filtering mechanism required depending on the event - want alarms directly relevant - not the cascade.
PM & Alarms c/o Niall
Get from logging system on demand rather than pull to files and store with PM data.
Adriaan Rijllart present the facilities that will be available for analysis of quenches etc. during hardware commissioning. At present the system doesn't interface to SDDS but this coming.
His presentation here.
Reported by Mike Lamont