[Rivet] seg fault in rivet-bootstrap-lcg in lxplus install

Andy Buckley andy.buckley at cern.ch
Wed Aug 3 11:15:31 BST 2016


What happens if you unset/don't set RIVET_ANALYSIS_PATH for now? If 
you've upgraded Rivet to a binary-incompatible version but not 
recompiled your private analyses, you could get segfaults.

In general an upgrade will require rebuilding private analyses -- we're 
not sophisticated enough to protect binary compatibility, do 
library-level version detection, etc. -- we'd need a lot more people to 
get that far down the priority list!

Andy


On 03/08/16 11:10, Raghav Kunnawalkam Elayavalli wrote:
> Andy,
> Ok something is very fishy here. this is my .bashrc
>
> # .bashrc
>
> # Source global definitions
> if [ -f /etc/bashrc ]; then
> . /etc/bashrc
> fi
> # source scripts to run cmssw commands and eos directories
> source /afs/cern.ch/cms/sw/cmsset_default.sh
> <http://cern.ch/cms/sw/cmsset_default.sh>
> # the above doesnt do anything since it doesnt exist.
> source /afs/cern.ch/project/eos/installation/user/etc/setup.sh
> <http://cern.ch/project/eos/installation/user/etc/setup.sh>
> #export EOS_MGM_URL=root://eos.cern.ch
> export EOS_MGM_URL=root://eosuser.cern.ch
> eosmount $HOME/eos
>
> export
> PATH=/afs/cern.ch/work/r/rkunnawa/JEWEL/jewel-2.1.0/local/bin:$PATH
> <http://cern.ch/work/r/rkunnawa/JEWEL/jewel-2.1.0/local/bin:$PATH>
> export
> LD_LIBRARY_PATH=/afs/cern.ch/work/r/rkunnawa/JEWEL/jewel-2.1.0/local/lib:$LD_LIBRARY_PATH
> <http://cern.ch/work/r/rkunnawa/JEWEL/jewel-2.1.0/local/lib:$LD_LIBRARY_PATH>
>
> export
> RIVET_ANALYSIS_PATH=/afs/cern.ch/work/r/rkunnawa/JEWEL/ANALYSIS/:$RIVET_ANALYSIS_PATH
> <http://cern.ch/work/r/rkunnawa/JEWEL/ANALYSIS/:$RIVET_ANALYSIS_PATH>
> export
> RIVET_ANALYSIS_PATH=/afs/cern.ch/work/r/rkunnawa/JEWEL/ANALYSIS/Validation/:$RIVET_ANALYSIS_PATH
> <http://cern.ch/work/r/rkunnawa/JEWEL/ANALYSIS/Validation/:$RIVET_ANALYSIS_PATH>
> export
> RIVET_ANALYSIS_PATH=/afs/cern.ch/work/r/rkunnawa/JEWEL/code:$RIVET_ANALYSIS_PATH
> <http://cern.ch/work/r/rkunnawa/JEWEL/code:$RIVET_ANALYSIS_PATH>
>
> and on top of this i just run the latest bootstrap_lcg and im still
> getting the seg fault. I tested in a new lxplus machine with no python
> path setup etc...
>
> Holger,
> hmm i did the diff, but i dont see anything obviously different. maybe
> im not looking at the right lines.
>
> Cheers
> Raghav
>> On Aug 3, 2016, at 11:55 AM, Holger Schulz <holger.schulz at durham.ac.uk
>> <mailto:holger.schulz at durham.ac.uk>> wrote:
>>
>> Weird.
>>
>> For me it works. I dumped my environment (after he source commands) here:
>>
>> https://users.hepforge.org/~holsch/Temp/env.dump
>>
>> Could you please run a diff against your env?
>>
>> Holger
>>
>>
>> On 03/08/16 10:34, Raghav Kunnawalkam Elayavalli wrote:
>>> just to make clear, i didnt have any PYTHONPATH set before i did this
>>> and the only other LD_LIBRARY_PATH was pointing to my MC generator
>>> path.
>>>
>>> Cheers
>>> Raghav
>>>> On Aug 3, 2016, at 11:32 AM, Raghav Kunnawalkam Elayavalli
>>>> <raghav.k.e at cern.ch <mailto:raghav.k.e at cern.ch>> wrote:
>>>>
>>>> Hi Holger,
>>>>
>>>> I also tried this morning with the same seg fault
>>>>
>>>> kunnawa:~ raghav$ cern
>>>> Password:
>>>> Last login: Tue Jun 21 08:56:17 2016 from kunnawa.dyndns.cern.ch
>>>> <http://kunnawa.dyndns.cern.ch/>
>>>> * ********************************************************************
>>>> * Welcome to lxplus105.cern.ch <http://lxplus105.cern.ch/>, SLC, 6.8
>>>> * Archive of news is available in /etc/motd-archive
>>>> * Reminder: You have agreed to comply with the CERN computing rules
>>>> * https://cern.ch/ComputingRules
>>>> * Puppet environment: production, Roger state: production
>>>> * Foreman hostgroup: bi/inter/plus/login
>>>> * LXPLUS Public Login Service
>>>> * ********************************************************************
>>>> -bash: /afs/cern.ch/cms/sw/cmsset_default.sh:
>>>> <http://cern.ch/cms/sw/cmsset_default.sh:> No such file or directory
>>>> OK
>>>> ===> Mountpoint   : /afs/cern.ch/user/r/rkunnawa/eos
>>>> <http://cern.ch/user/r/rkunnawa/eos>
>>>> ===> Fuse-Options :
>>>> kernel_cache,attr_timeout=30,entry_timeout=30,max_readahead=131072,max_write=4194304,fsname=eosuser.cern.ch
>>>> <http://eosuser.cern.ch/> root://eosuser.cern.ch//eos/
>>>> ===> xrootd ra             : 131072
>>>> ===> xrootd cache          : 393216
>>>> ===> fuse debug            : 0
>>>> ===> fuse write-cache      : 1
>>>> ===> fuse write-cache-size : 100000000
>>>> [rkunnawa at lxplus105 ~]$ work
>>>> [rkunnawa at lxplus105 rkunnawa]$ mkdir RIVET_test
>>>> [rkunnawa at lxplus105 rkunnawa]$ cd RIVET_test/
>>>> [rkunnawa at lxplus105 RIVET_test]$ ls
>>>> [rkunnawa at lxplus105 RIVET_test]$ source
>>>> /afs/cern.ch/sw/lcg/releases/LCG_84/Python/2.7.10/x86_64-slc6-gcc48-opt/Python-env.sh
>>>> <http://cern.ch/sw/lcg/releases/LCG_84/Python/2.7.10/x86_64-slc6-gcc48-opt/Python-env.sh>
>>>> [rkunnawa at lxplus105 RIVET_test]$ source
>>>> /afs/cern.ch/sw/lcg/releases/LCG_84/MCGenerators/rivet/2.5.0/x86_64-slc6-gcc48-opt/rivetenv.sh
>>>> <http://cern.ch/sw/lcg/releases/LCG_84/MCGenerators/rivet/2.5.0/x86_64-slc6-gcc48-opt/rivetenv.sh>
>>>> [rkunnawa at lxplus105 RIVET_test]$ which rivet
>>>> /afs/cern.ch/sw/lcg/releases/MCGenerators/rivet/2.5.0-1aa46/x86_64-slc6-gcc48-opt/bin/rivet
>>>> <http://cern.ch/sw/lcg/releases/MCGenerators/rivet/2.5.0-1aa46/x86_64-slc6-gcc48-opt/bin/rivet>
>>>> [rkunnawa at lxplus105 RIVET_test]$ rivet --list-analyses
>>>> Segmentation fault
>>>>
>>>> Cheers
>>>> Raghav Kunnawalkam Elayavalli
>>>> PhD Candiate
>>>> Rutgers University
>>>> http://physics.rutgers.edu/~raghav/
>>>> <http://www.google.com/url?q=http%3A%2F%2Fphysics.rutgers.edu%2F%7Eraghav%2F&sa=D&sntz=1&usg=AFQjCNENvwedEL22DMATPxyb0WMvVEm9Bw>
>>>>
>>>>> On Aug 3, 2016, at 11:30 AM, Holger Schulz
>>>>> <holger.schulz at durham.ac.uk <mailto:holger.schulz at durham.ac.uk>> wrote:
>>>>>
>>>>> Hi Raghav,
>>>>>
>>>>> we recently held a Rivet tutorial at CERN using prebuilt rivet2.5
>>>>> from LCG:
>>>>>
>>>>> source
>>>>> /afs/cern.ch/sw/lcg/releases/LCG_84/Python/2.7.10/x86_64-slc6-gcc48-opt/Python-env.sh
>>>>> <http://cern.ch/sw/lcg/releases/LCG_84/Python/2.7.10/x86_64-slc6-gcc48-opt/Python-env.sh>
>>>>> source
>>>>> /afs/cern.ch/sw/lcg/releases/LCG_84/MCGenerators/rivet/2.5.0/x86_64-slc6-gcc48-opt/rivetenv.sh
>>>>> <http://cern.ch/sw/lcg/releases/LCG_84/MCGenerators/rivet/2.5.0/x86_64-slc6-gcc48-opt/rivetenv.sh>
>>>>>
>>>>> That worked fine for people from CMS, ATLAS and ALICE. There were a
>>>>> couple of people who had
>>>>> to pruge their bashrc becaus of conflicting PYTHONPATH,
>>>>> LD_LIBRARY_PATH things that
>>>>> typically had to do with experiment specific setups of python.
>>>>>
>>>>> Why don't you give that a try first?
>>>>>
>>>>> Thanks,
>>>>> Holger
>>>>>
>>>>>
>>>>> On 03/08/16 10:23, Andy Buckley wrote:
>>>>>> Hi Raghav,
>>>>>>
>>>>>> This is very strange. I just tried the 2.5.0 version of
>>>>>> rivet-bootstrap-lcg and it worked fine, giving me a functioning
>>>>>> setup. At least it could list analyses without segfaulting.
>>>>>>
>>>>>> It's possible that I've got something set by default in my shell
>>>>>> environment that helps me, but by default on lxplus I get the same
>>>>>> gcc 4.4 and Python 2.6 as you do. Maybe someone else could try to
>>>>>> reproduce your problem.
>>>>>>
>>>>>> I've just updated the LCG bootstrap script to use LCG_85, a newer
>>>>>> GCC and FastJet, etc. This also works for me. Maybe you'd have
>>>>>> more luck with that?
>>>>>>
>>>>>> To use the normal bootstrap on lxplus and other SL6 systems with
>>>>>> an old system compiler you need to set up your environment to use
>>>>>> a newer compiler and a version of Python compiled with the same.
>>>>>> The LCG bootstrap does this automatically and puts the relevant
>>>>>> lines into the env.sh script. From my updated LCG_85 build, the
>>>>>> relevant lines are these:
>>>>>>
>>>>>> source
>>>>>> /cvmfs/sft.cern.ch/lcg/releases/LCG_85/gcc/4.9.3/x86_64-slc6/setup.sh
>>>>>> <http://sft.cern.ch/lcg/releases/LCG_85/gcc/4.9.3/x86_64-slc6/setup.sh>
>>>>>> source
>>>>>> /cvmfs/sft.cern.ch/lcg/releases/LCG_85/Python/2.7.10/x86_64-slc6-gcc49-opt/Python-env.sh
>>>>>> <http://sft.cern.ch/lcg/releases/LCG_85/Python/2.7.10/x86_64-slc6-gcc49-opt/Python-env.sh>
>>>>>>
>>>>>> There are other, fancier ways to do it, but hopefully this will
>>>>>> help. It's definitely a good idea to use a newer release than
>>>>>> 2.4.1 -- at least 2.4.3, please! We will soon be releasing 2.5.1.
>>>>>>
>>>>>> Andy
>>>>>>
>>>>>>
>>>>>>
>>>>>> On 02/08/16 16:23, Raghav Kunnawalkam Elayavalli wrote:
>>>>>>> Hi Rivet Experts,
>>>>>>>
>>>>>>> I tried a fresh install of the rivet-bootstrap-lcg out of the box
>>>>>>> in an
>>>>>>> lxplus machine and after i did
>>>>>>> $ ./rivet-bootstrap-lcg
>>>>>>> [rkunnawa at lxplus016 RIVET]$ source
>>>>>>> /afs/cern.ch/work/r/rkunnawa/RIVET/local/env.sh
>>>>>>> <http://cern.ch/work/r/rkunnawa/RIVET/local/env.sh>
>>>>>>> <http://cern.ch/work/r/rkunnawa/RIVET/local/env.sh>
>>>>>>> [rkunnawa at lxplus016 RIVET]$ which rivet
>>>>>>> /afs/cern.ch/work/r/rkunnawa/RIVET/local/bin/rivet
>>>>>>> <http://cern.ch/work/r/rkunnawa/RIVET/local/bin/rivet>
>>>>>>> <http://cern.ch/work/r/rkunnawa/RIVET/local/bin/rivet>
>>>>>>> [rkunnawa at lxplus016 RIVET]$ which python
>>>>>>> /cvmfs/sft.cern.ch/lcg/releases/LCG_82/Python/2.7.10/x86_64-slc6-gcc48-opt/bin/python
>>>>>>> <http://sft.cern.ch/lcg/releases/LCG_82/Python/2.7.10/x86_64-slc6-gcc48-opt/bin/python>
>>>>>>>
>>>>>>> <http://sft.cern.ch/lcg/releases/LCG_82/Python/2.7.10/x86_64-slc6-gcc48-opt/bin/python>
>>>>>>>
>>>>>>> [rkunnawa at lxplus016 RIVET]$ echo $PYTHONPATH
>>>>>>> /afs/cern.ch/work/r/rkunnawa/RIVET/local/lib/python2.7/site-packages:/afs/cern.ch/work/r/rkunnawa/RIVET/local/lib/python2.7/site-packages:/afs/cern.ch/work/r/rkunnawa/RIVET/local/lib/python2.7/site-packages:/cvmfs/sft.cern.ch/lcg/releases/LCG_82/Python/2.7.10/x86_64-slc6-gcc48-opt/lib/python2.7/site-packages:/afs/cern.ch/work/r/rkunnawa/RIVET/local/lib/python2.7/site-packages:/afs/cern.ch/work/r/rkunnawa/RIVET/local/lib/python2.7/site-packages:/afs/cern.ch/work/r/rkunnawa/RIVET/local/lib/python2.7/site-packages:/cvmfs/sft.cern.ch/lcg/releases/LCG_85/Python/2.7.10/x86_64-slc6-gcc49-opt/lib/python2.7/site-packages:
>>>>>>> <http://cern.ch/work/r/rkunnawa/RIVET/local/lib/python2.7/site-packages:/afs/cern.ch/work/r/rkunnawa/RIVET/local/lib/python2.7/site-packages:/afs/cern.ch/work/r/rkunnawa/RIVET/local/lib/python2.7/site-packages:/cvmfs/sft.cern.ch/lcg/releases/LCG_82/Python/2.7.10/x86_64-slc6-gcc48-opt/lib/python2.7/site-packages:/afs/cern.ch/work/r/rkunnawa/RIVET/local/lib/python2.7/site-packages:/afs/cern.ch/work/r/rkunnawa/RIVET/local/lib/python2.7/site-packages:/afs/cern.ch/work/r/rkunnawa/RIVET/local/lib/python2.7/site-packages:/cvmfs/sft.cern.ch/lcg/releases/LCG_85/Python/2.7.10/x86_64-slc6-gcc49-opt/lib/python2.7/site-packages:>
>>>>>>>
>>>>>>> <http://cern.ch/work/r/rkunnawa/RIVET/local/lib/python2.7/site-packages:/afs/cern.ch/work/r/rkunnawa/RIVET/local/lib/python2.7/site-packages:/afs/cern.ch/work/r/rkunnawa/RIVET/local/lib/python2.7/site-packages:/cvmfs/sft.cern.ch/lcg/releases/LCG_82/Python/2.7.10/x86_64-slc6-gcc48-opt/lib/python2.7/site-packages:/afs/cern.ch/work/r/rkunnawa/RIVET/local/lib/python2.7/site-packages:/afs/cern.ch/work/r/rkunnawa/RIVET/local/lib/python2.7/site-packages:/afs/cern.ch/work/r/rkunnawa/RIVET/local/lib/python2.7/site-packages:/cvmfs/sft.cern.ch/lcg/releases/LCG_85/Python/2.7.10/x86_64-slc6-gcc49-opt/lib/python2.7/site-packages:>
>>>>>>>
>>>>>>> [rkunnawa at lxplus016 RIVET]$ rivet --list-analyses
>>>>>>> Segmentation fault
>>>>>>> [rkunnawa at lxplus016 RIVET]$
>>>>>>>
>>>>>>> I tried this with LCG_82 and LCG_85 (with appropriately changing the
>>>>>>> other locations) and i get the same issues. any ideas whats
>>>>>>> happening?
>>>>>>>
>>>>>>> I also tried to install rivet with the normal bootstrap and i got an
>>>>>>> issue with yoda installation since the default gcc in lxplus is
>>>>>>> 4.4 and
>>>>>>> it doesnt support C++11. so any ideas how i can over come that
>>>>>>> issue? I
>>>>>>> would really like to use something like the herwig bootstrap which
>>>>>>> installs all the necessary items.
>>>>>>>
>>>>>>> In the meanwhile should i try to install an older version of
>>>>>>> rivet? one
>>>>>>> that doesnt need c++11?
>>>>>>> Again, the whole reason why im using this is so that i can run my
>>>>>>> rivet
>>>>>>> analysis on the lxplus batch system since at the moment (my
>>>>>>> earlier lcg
>>>>>>> installation) doesnt work and gives me weird python errors which even
>>>>>>> the cern service desk is having a hard time with.
>>>>>>>
>>>>>>> Cheers
>>>>>>> Raghav Kunnawalkam Elayavalli
>>>>>>> PhD Candiate
>>>>>>> Rutgers University
>>>>>>> http://physics.rutgers.edu/~raghav/
>>>>>>> <http://physics.rutgers.edu/%7Eraghav/>
>>>>>>> <http://www.google.com/url?q=http%3A%2F%2Fphysics.rutgers.edu%2F~raghav%2F&sa=D&sntz=1&usg=AFQjCNENvwedEL22DMATPxyb0WMvVEm9Bw
>>>>>>> <http://www.google.com/url?q=http%3A%2F%2Fphysics.rutgers.edu%2F%7Eraghav%2F&sa=D&sntz=1&usg=AFQjCNENvwedEL22DMATPxyb0WMvVEm9Bw>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> _______________________________________________
>>>>>>> Rivet mailing list
>>>>>>> Rivet at projects.hepforge.org <mailto:Rivet at projects.hepforge.org>
>>>>>>> https://www.hepforge.org/lists/listinfo/rivet
>>>>>>>
>>>>>>
>>>>>>
>>>>>
>>>>
>>>
>>
>


-- 
Dr Andy Buckley, Lecturer / Royal Society University Research Fellow
Particle Physics Expt Group, University of Glasgow


More information about the Rivet mailing list