JCOP FWWG 14 February 2017

→ Europe/Zurich
4/S-030 (CERN)

4/S-030

CERN

30
Show room on map
Description
The meeting dedicated to CMS's extensions of fwInstallation and fwConfigurationDB, as well as discussions of fwInstallation tool development progress and plans

1) General News

- Beta release of JCOP Framework 8.0.3 available. Will be released as final if no bugs found by the end of the week

- A new CtrlRDBAccess version will be available shortly resolving the issues found in CMS during component installation.

2) CMS ECAL extension to the fwInstallation mechanism

- fwInstallationUtils is a library of functions that could simplify the postInstallation script. They are driven with a XML file to standardize operations on managers (start/stop simulators before devices are created, etc).

- no much feedback from other experiments, as they are not instantiating devices this way (they do not package complete apps as components); there are a few application in the accelerator sector that seem to be doing a similar way (eg. PIC), yet vast majority rely on UNICOS-imports done manually, ie. "online engineering"

- enforcing/validating configuration with this component is probably not of interest as this is done already by system integrity

3) CMS extensions of fwConfigurationDB

- allows for easier selection of devices and datapoints, and to save hw and logical view at once

- has a "comparison" feature for addresses, archiving and alerts (live system against what is in DB); feature could be factorized out, as there are no much dependencies on CMS-specific code.

- question about performance of comparison: ~3 minutes for 4000 datapoints

- possible merger of the tool to ConfigurationDB? Not much interest from other experiments, as they are not using ConfDB to store static configurations. Some part of device selector would, however, be interesting

4) Update on the status of fwInstallation

- some fixes in DB Schema (notably for ALICE and CMS) as well as in libs

- ASCII imports are now twice faster on Windows. Enabled batching (on Windows by default it is set to 1).

- Should we consider merging all DPL files of all components before installation?
=> not sure about the actual gain, and debugging becomes more cumbersome; moreover, if there are many DPEs being ASCII-imported at once the DM becomes very busy, hence maybe small files are actually optimal

- priorities for further work to be defined. A working group (one person per experiment nominated by the Controls Coordinator) will likely be spawned.

- there seems to be a memory leak in the fwInstallation (from CtrlRDBAccess) in ALICE. Lukasz looking into this. The new CtrlRDBAccess that is going to come shortly in the very next CERN patch may actually already address this

- similar memory-leak issues seem to be visible in LHCb yet require more pinpointing. Clara will check and contact Piotr

- Is it possible to target a component to a specific WinCC OA version?
Yes, there is a dedicated tag for this in the component XML fle (<required_pvss_version>), and it is used e.g. in the Core component and others. To follow up is how the Central Installation would handle it.

4) Status on WinCC OA upgrades

 - ~150 systems in BE/ICS

- LHCB: 160 projects completes in 2 days

- CMS: all upgraded except 3, out of which 2 postponed

- ALICE: Central systems done, and 2 out of 20 subdetectors

There are minutes attached to this event. Show them.