Andrea Hawkins wrote:
> In response to John's first thought:
> 
>  grep uuid $PETSC_DIR/bmake/$PETSC_ARCH/petscconf
> MPI_LIB = 
> -Wl,-rpath,/org/groups/oden/LIBRARIES/MPI/mpich-ahawkins/mpich2-1.0.8-intel-shared-dbg/lib
> -L/org/groups/oden/LIBRARIES/MPI/mpich-ahawkins/mpich2-1.0.8-intel-shared-dbg/lib
> -lmpich -lssl -luuid -lpthread -lrt -ldl
> 
> So, I guess it must be something set for MPI and/or PETSc.
> 
> I've been trying some other configure/compiler combinations since my
> e-mail, though, so I don't think make echo will work...
> 
> My current make is for a run using gcc, and make echo doesn't seem to
> have luuid in it. So maybe it is something with the intel compiler.

Looks like it's something with mpich2 (which wants to use libuuid to 
generate globally unique ids) and with ICES (which currently only has 
/lib/libuuid.so.1 and /lib/libuuid.so.1.2, not the /lib/libuuid.so that 
that linker statement wants to find).  Reconfiguring and rebuilding 
mpich2 might fix it; so might hassling the ICES sysadmins.

There's been some weirdness with ICES software upgrades during the past 
few weeks, I think; I'm still trying to figure out why my intel-based 
builds are working fine why my custom-gcc-based builds have started 
giving "/usr/lib/libstdc++.so.6: version `GLIBCXX_3.4.10' not found" and
"/usr/lib/libstdc++.so.6: version `CXXABI_1.3.2' not found" errors.
---
Roy

------------------------------------------------------------------------------
This SF.net email is sponsored by:
SourcForge Community
SourceForge wants to tell your story.
http://p.sf.net/sfu/sf-spreadtheword
_______________________________________________
Libmesh-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/libmesh-users

Reply via email to