[FLASH-USERS] Compilation error with Parallel HDF5 1.10.5

[Mattia Cipriani]

Dear Klaus and the others who helped me,

I found the solution to the last problem: it was due to an error in the LD_LIBRARY_PATH in my ~/.profile file.
Since I was appending to the LD_LIBRARY_PATH the path to a serial hdf5 build, the linker was looking in the wrong path, because the good one was not written correctly. When I deleted the non-parallel path, the linker was not able to find the hdf5 library at all, and I was able to understand the mistake.
The make process for FLASH instead was successful because the path to the hdf5 library was correct in the Makefile.h.
This made the debug tricky.

Thank you again, 
Best
Mattia

························································
  Dr Mattia Cipriani, PhD
    FSN-FUSPHY-SAD
    ENEA Centro Ricerche Frascati
    45, Via E. Fermi
    00044 Frascati (Rome)
    Italy
    Ufficio: Edificio 12
                Stanza 12029
    Tel.      +39 069400 5145 
    mobile +39 328 5552077
························································




> Il giorno 11 mar 2019, alle ore 20:51, Mattia Cipriani <mattia.cipriani at enea.it> ha scritto:
> 
> Dear Klaus,
> 
> Yes, it is surprising because the hdf5 library has been built in parallel and also the tests made using make check were all good.
> I understand the difficulty of debugging remotely, I really appreciated your help.
> I will try by myself to understand if the problem is related to the system configuration. Since I am running flash on a cluster it might be the case.
> 
> Best regards,
> Mattia
> 
> ························································
>   Dr Mattia Cipriani, PhD
>     FSN-FUSPHY-SAD
>     ENEA Centro Ricerche Frascati
>     45, Via E. Fermi <x-apple-data-detectors://2>
>     00044 <x-apple-data-detectors://2> Frascati (Rome)
>     Italy
>     Ufficio: Edificio 12
>                 Stanza 12029
>     Tel.      +39 069400 5145 <tel:+39%20069400%205145> 
>     mobile +39 328 5552077 <tel:+39%20328%205552077>
> ························································
> 
> Il giorno 11 mar 2019, alle ore 17:58, Klaus Weide <klaus at flash.uchicago.edu <mailto:klaus at flash.uchicago.edu>> ha scritto:
> 
>> On Sat, 9 Mar 2019, Mattia Cipriani wrote:
>> 
>>> Dear Klaus,
>>> 
>>> thank you very much for the patch, which allowed to reach the end of the compilation.
>>> However now, when I run the code, I get the following error, at the moment of outputting the first hdf5 file:
>>> flash4: symbol lookup error: flash4: undefined symbol: H5Pset_fapl_mpio
>> 
>> I think this indicates that the HDF5 library with which you have linked 
>> FLASH has not been built with parallel support - although it is surprising 
>> that you did not already get an error at link time (i.e., in the last 
>> stage of running 'make').
>> 
>> There may be additional sources of problems if you are running flash4 in a 
>> different environment from that in which it was built - in particular, if 
>> you are running under a batch system - relating to locations of 
>> dynamically linked libraries, LD_LIBRARY_PATH, etc.
>> 
>> This is hard to impossible to diagnose remotely without access to the 
>> system.
>> 
>> Klaus

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://flash.rochester.edu/pipermail/flash-users/attachments/20190312/0bcc57cf/attachment.htm>