LHCb Core Software Meeting
Date and Location
19 Mar 2014
10:30 - 11:05
CERN (2-R-030)
Attendees
Anastasia Karachaliou,
Ben Couturier,
Illya Shapoval,
Joel Closier,
Liang Sun,
Marco Cattaneo,
Marco Clemencic (minutes),
Markus Frank,
Sebastien Binet (Vidyo)
Apologies
Gloria Corti, Rob Lambert
Subjects
Round Table
Rob (email)
- https://savannah.cern.ch/task/?49868
: I wanted to mention the gaudiexcise.py script I have worked on, which I think should migrate to lbscripts. It passes all arguments transparently to gaudirun.py aside from the first argument which is the name of an algorithm to excise from gaudirun, making a sandbox in which that algorithm can run stand-alone. I'm going to write a twiki page for this, but it's not quite ready yet.
- https://savannah.cern.ch/task/?49844
: I also wanted to mention the new LHCbTimingAuditor which can output to a file instead of stdout, the three classes, LHCbTimerForSequencer, LHCbSequencerTimerTool and LHCbTimingAuditor, I've modified, are only very superficial edits of the existing Gaudi classes, and if you renamed them back they could replace those in Gaudi for the next Gaudi release. They only add functionality and flexibility.
- Then I wanted to remind about the getNightlyRefs.py script which copies reference files from the results of the nightly builds. Since I am leaving this should get copied somewhere other than my public area, and Lbscripts seems the logical place, although probably the script could be tidied up a bit and homogenized to follow the convention of the other lbscripts, but maybe that's not worth the effort since so few people will be using the script in the end. Currently it's in ~rlambert/public/forMarco/getNightlyRefs.py
Marco Cl.
- CMake
- initial migration of Orwell and Vetra, waiting for the nightlies to validate the builds
- ROOT 6
- a usable version has been tagged on Friday, waiting for the new SFT deployment machinery
- LCGCMT
- still validating the new deployment machinery (several iterations so far)
- Gaudi
- to be tagged once we have the new LCG_68 versions working (with the new deployment machinery)
- write access to the repository (and instructions) given to Ben and Sebastien (as back up)
Marco Ca.
- Tagged the whole stack up to Brunel, waiting to upgrade the Gaudi version
Illya
- Ariadne: Fighting with the consequences of the unclean shutdown of Neo4J.
- cleaned up (partially) the node is space (with the help from the Neo4J support)
- would be nice to have the alarms
Marco Cl.: we need to schedule the meeting on monitoring and alarms
Ben: need help to check the deployment database
Joel
- Federico proposed some way of sharing the list of platform tags etc.
- Ben
- We need some more discussion
Marco Cl.: Stefan should check that everywhere on the grid we have SLC 6.5 (issue with libssl version between 6.4 and 6.5)
Joel: it is now possible to create SLC 6.4 virtual machines, so Benedikt can extend his pool
Ben
- Discussion on going with Stefan Roiser, who wants to use only CVMFS for the software deployment.
- there is a general consensus that this cannot work for all the use cases
Liang
- Progressing with CondDb for split HLT
Sebastien
Marco Cl.: we need to discuss with Benedikt a way to access tools like clang-modernizer etc. (wrapper scripts?)
Actions
- Set up a meeting to discuss monitoring and alarms Marco Cl.
- Discuss with SFT a way of distributing various tools, like clang-format Marco Cl., Ben
- Check the projects deployment db, and migrate it to the official Neo4J instance Ben, Illya
- Once ready, move gaudiexcise.py to either Gaudi or GaudiConf Marco Cl., Rob LBCORE-426
- Once stable, move the LHCbTimingAuditor to Gaudi Ben, Rob LBCORE-426
- Move getNightlyRefs.py to LbScripts, possibly renaming it using the new policy (lb-...) Ben LBCORE-426
--
MarcoClemencic - 19 Mar 2014