[Users] Compiling Einstein Toolkit

Hamilton, Maria babiuc at marshall.edu
Wed Jul 27 14:01:18 CDT 2016


Roland,

I’m coming back to the problem of /usr/bin/ld: cannot find –lz. 
I looked at configs/ET-config/config-info and configs/ET-config/config-data/make.config.defn and did not find the offending lz or ld. 
I also tried to load the hdf5 module and give the directory instead of building it. 

Rahul, 

I’m inclined to believe the problem lies with szip. I asked the person who manages the cluster, and found that the hdf5 and hdf4 modules are ‘hidden’ by the SDSC HDF roll’s environment module settings and require the ‘gnu’ compiler environment module before module avail will show them. 
But module load gnu hdf5 and then rebuild did not fix it either. 
Is there a way to maybe give the path directly in the cfg file, by means of LIBDIRS =?
I see /lib64/libz.so.1  and /lib64/libz.so.1.2.3. Shall I use one of them, or ask the lcluster manager to install szip as a module ?

Thanks,
Maria

On 6/29/16, 2:54 PM, "users-bounces at einsteintoolkit.org on behalf of Roland Haas" <users-bounces at einsteintoolkit.org on behalf of rhaas at aei.mpg.de> wrote:

    Hello Maria,
    
    I belietve this could also be caused by incorrect LIBS settings.
    
    For this, in your option list, please check if you added something like
    "-lz" to the one of the LIBS variables. It should really read just "z".
    
    You should be able to check what is used (but not modify it easily) by
    looking at configs/ET-config/config-info or if you need the real thing
    then configs/ET-config/config-data/make.config.defn then you can modify
    simfactory/mdb/optionlists/biggreen.cfg accordingly.
    
    You may also want to double check that you did not accidentally add
    something like LD=ld to your option list which is no correct (it should
    be g++).
    
    Admittedly the biggreen.cfg file that you had included a while back
    does not seem to do anything like this.
    
    One more thing: looking at the cfg file, I believe that it should read
    MPI_DIR=/opt/openmpi rather than OPENMPI_DIR=/opt/openmpi
    
    Yours,
    Roland
    
    > Hello,
    > 
    > Indeed, gcc was outdated. I loaded gnu/4.9.2.
    > I got it configured, but at make I have this error:
    > /usr/bin/ld: cannot find -lz
    > collect2: error: ld returned 1 exit status
    > 
    > What is your suggestion?
    > 
    > Thank you for your help.
    > Maria
    > 
    > On 6/28/16, 5:14 PM, "Korobkin, Oleg" <korobkin at lanl.gov> wrote:
    > 
    > Maria,
    > 
    > Looks like default g++ on your cluster is outdated. You can check the
    > version by typing,
    > $ g++ ‹-version
    > 
    > Versions before 4.7.0 don¹t support the offending '‹std=gnu++11'
    > flag. But the cluster certainly should have more recent compilers
    > installed. Is there a link to the list of software available on
    > biggreen? Try this command:
    > $ module avail
    > 
    >  - this should list all the modules that are installed on your
    > cluster, including more recent compilers. If any modules are present,
    > they are probably called 'gcc/X.Y.Z' or 'gnu-compilers-X.Y.Z' and you
    > can add the corresponding module like this:
    > $ module load gcc/4.8.1
    > 
    > Try adding this newer module and recompiling:
    > $ make ET-config options=simfactory/mdb/optionlists/biggreen.cfg
    > THORNLIST=manifest/einsteintoolkit.th
    > 
    > 
    > All the best,
    >  - Oleg
    > 
    > On 6/28/16, 11:59 AM, "users-bounces at einsteintoolkit.org on behalf of
    > Hamilton, Maria" <users-bounces at einsteintoolkit.org on behalf of
    > babiuc at marshall.edu> wrote:  
    > 
    > >Hello,
    > >
    > >I was wondering if I can get some help in compiling the Einstein
    > >Toolkit on a cluster called BigGreen
    > >http://www.marshall.edu/itresearch/resources-and-tools/biggreen-cluster/
    > >I¹ve followed the instructions in
    > >https://docs.einsteintoolkit.org/et-docs/Compiling_the_Einstein_Toolkit
    > >I¹ve created biggreen.marshall.edu.ini and biggreen.cfg, which are
    > >attached to this message.
    > >When I issued the commands:
    > >simfactory/bin/sim setup --optionlist= biggreen.cfg --runscript
    > >biggreen.sh
    > >simfactory/bin/sim build  --thornlist=manifest/einsteintoolkit.th
    > >I get this error:
    > >Cactus requires a C++11 compiler -- check your C++ compiler and C++
    > >compiler flags
    > >
    > >When I run:
    > >make ET-config options=simfactory/mdb/optionlists/biggreen.cfg
    > >THORNLIST=manifest/einsteintoolkit.th
    > >I get this error:
    > >checking whether the C++ compiler (g++ -g3 -march=native -std=gnu++11
    > >-rdynamic) works... no
    > >configure: error: installation or configuration problem: C++ compiler
    > >cannot create executables (see
    > >configs/<configname>/config-data/config.log for details).
    > >
    > >I am attaching also the config.log.
    > >
    > >Thank you,
    > >Maria Babiuc
    > >  
    > 
    > 
    > 
    > _______________________________________________
    > Users mailing list
    > Users at einsteintoolkit.org
    > http://lists.einsteintoolkit.org/mailman/listinfo/users
    
    
    
    -- 
    My email is as private as my paper mail. I therefore support encrypting
    and signing email messages. Get my PGP key from http://keys.gnupg.net.
    



More information about the Users mailing list