Wow - strange. How was this configured?

Something clearly is messed up as there is no such thing as a "basic" grpcomm module, or an ess "slurmd" module, etc. any more. My best guess is that you wound up installing this on top of an existing OMPI install, and have now created a massive library confusion problem.

Try configuring it --with-prefix=<foo>, where foo is under your home directory. This will provide a clean place to install without confusion. Then put that location at the front of your path and ld_library_path - I suspect things will be okay from there.


On Dec 23, 2012, at 1:43 PM, R C <recifx@gmail.com> wrote:

I put /usr/local/lib at the beginning of LD_LIBRARY_PATH and I get:
$ ompi_info
[RCAsus1:20105] mca: base: component_find: unable to open /usr/local/lib/openmpi/mca_errmgr_default: /usr/local/lib/openmpi/mca_errmgr_default.so: undefined symbol: orte_errmgr_base_output (ignored)
[RCAsus1:20105] mca: base: component_find: unable to open /usr/local/lib/openmpi/mca_ess_slurmd: /usr/local/lib/openmpi/mca_ess_slurmd.so: undefined symbol: orte_nid_t_class (ignored)
[RCAsus1:20105] mca: base: component_find: unable to open /usr/local/lib/openmpi/mca_grpcomm_basic: /usr/local/lib/openmpi/mca_grpcomm_basic.so: undefined symbol: opal_profile_file (ignored)
[RCAsus1:20105] mca: base: component_find: unable to open /usr/local/lib/openmpi/mca_routed_cm: /usr/local/lib/openmpi/mca_routed_cm.so: undefined symbol: orte_message_event_t_class (ignored)
[RCAsus1:20105] mca: base: component_find: unable to open /usr/local/lib/openmpi/mca_routed_linear: /usr/local/lib/openmpi/mca_routed_linear.so: undefined symbol: orte_message_event_t_class (ignored)
                 Package: Open MPI rc@RCAsus1 Distribution
                Open MPI: 1.9a1r27719
  Open MPI repo revision: r27719
   Open MPI release date: Unreleased developer copy
                Open RTE: 1.9a1r27719


and now when I try:
$ mpicc -o hw hello.c
it seems to compile, but from:
$mpirun -np 4 hw
RCAsus1:26668] mca: base: component_find: unable to open /usr/local/lib/openmpi/mca_ess_slurmd: /usr/local/lib/openmpi/mca_ess_slurmd.so: undefined symbol: orte_nid_t_class (ignored)
[RCAsus1:26668] mca: base: component_find: unable to open /usr/local/lib/openmpi/mca_errmgr_default: /usr/local/lib/openmpi/mca_errmgr_default.so: undefined symbol: orte_errmgr_base_output (ignored)
[RCAsus1:26668] mca: base: component_find: unable to open /usr/local/lib/openmpi/mca_routed_cm: /usr/local/lib/openmpi/mca_routed_cm.so: undefined symbol: orte_message_event_t_class (ignored)
[RCAsus1:26668] mca: base: component_find: unable to open /usr/local/lib/openmpi/mca_routed_linear: /usr/local/lib/openmpi/mca_routed_linear.so: undefined symbol: orte_message_event_t_class (ignored)
[RCAsus1:26668] mca: base: component_find: unable to open /usr/local/lib/openmpi/mca_grpcomm_basic: /usr/local/lib/openmpi/mca_grpcomm_basic.so: undefined symbol: opal_profile_file (ignored)
[RCAsus1:26668] *** Process received signal ***
[RCAsus1:26668] Signal: Segmentation fault (11)
[RCAsus1:26668] Signal code:  (128)
[RCAsus1:26668] Failing at address: (nil)
[RCAsus1:26668] [ 0] /lib/x86_64-linux-gnu/libpthread.so.0(+0xfcb0) [0x7f0e31d54cb0]
[RCAsus1:26668] [ 1] /usr/local/lib/openmpi/mca_rmaps_load_balance.so(+0xca8) [0x7f0e2f720ca8]
[RCAsus1:26668] [ 2] /usr/local/lib/libopen-rte.so.0(orte_rmaps_base_map_job+0x1ba) [0x7f0e3226b09a]
[RCAsus1:26668] [ 3] /usr/local/lib/libopen-pal.so.0(opal_libevent2019_event_base_loop+0x626) [0x7f0e31fb97e6]
[RCAsus1:26668] [ 4] /usr/local/lib/libopen-pal.so.0(opal_progress+0xb3) [0x7f0e31f87483]
[RCAsus1:26668] [ 5] mpirun(orterun+0x10fb) [0x404b8d]
[RCAsus1:26668] [ 6] mpirun(main+0x20) [0x4036ec]
[RCAsus1:26668] [ 7] /lib/x86_64-linux-gnu/libc.so.6(__libc_start_main+0xed) [0x7f0e319a776d]
[RCAsus1:26668] [ 8] mpirun() [0x4035e9]
[RCAsus1:26668] *** End of error message ***
Segmentation fault (core dumped)



504:5 $ mpicc -o hw hello.c
[04:37 PM][rc@RCAsus1] [~/Dropbox/tst]
505:6 $ mpirun -np 4 hw
[RCAsus1:26668] mca: base: component_find: unable to open /usr/local/lib/openmpi/mca_ess_slurmd: /usr/local/lib/openmpi/mca_ess_slurmd.so: undefined symbol: orte_nid_t_class (ignored)
[RCAsus1:26668] mca: base: component_find: unable to open /usr/local/lib/openmpi/mca_errmgr_default: /usr/local/lib/openmpi/mca_errmgr_default.so: undefined symbol: orte_errmgr_base_output (ignored)
[RCAsus1:26668] mca: base: component_find: unable to open /usr/local/lib/openmpi/mca_routed_cm: /usr/local/lib/openmpi/mca_routed_cm.so: undefined symbol: orte_message_event_t_class (ignored)
[RCAsus1:26668] mca: base: component_find: unable to open /usr/local/lib/openmpi/mca_routed_linear: /usr/local/lib/openmpi/mca_routed_linear.so: undefined symbol: orte_message_event_t_class (ignored)
[RCAsus1:26668] mca: base: component_find: unable to open /usr/local/lib/openmpi/mca_grpcomm_basic: /usr/local/lib/openmpi/mca_grpcomm_basic.so: undefined symbol: opal_profile_file (ignored)
[RCAsus1:26668] *** Process received signal ***
[RCAsus1:26668] Signal: Segmentation fault (11)
[RCAsus1:26668] Signal code:  (128)
[RCAsus1:26668] Failing at address: (nil)
[RCAsus1:26668] [ 0] /lib/x86_64-linux-gnu/libpthread.so.0(+0xfcb0) [0x7f0e31d54cb0]
[RCAsus1:26668] [ 1] /usr/local/lib/openmpi/mca_rmaps_load_balance.so(+0xca8) [0x7f0e2f720ca8]
[RCAsus1:26668] [ 2] /usr/local/lib/libopen-rte.so.0(orte_rmaps_base_map_job+0x1ba) [0x7f0e3226b09a]
[RCAsus1:26668] [ 3] /usr/local/lib/libopen-pal.so.0(opal_libevent2019_event_base_loop+0x626) [0x7f0e31fb97e6]
[RCAsus1:26668] [ 4] /usr/local/lib/libopen-pal.so.0(opal_progress+0xb3) [0x7f0e31f87483]
[RCAsus1:26668] [ 5] mpirun(orterun+0x10fb) [0x404b8d]
[RCAsus1:26668] [ 6] mpirun(main+0x20) [0x4036ec]
[RCAsus1:26668] [ 7] /lib/x86_64-linux-gnu/libc.so.6(__libc_start_main+0xed) [0x7f0e319a776d]
[RCAsus1:26668] [ 8] mpirun() [0x4035e9]
[RCAsus1:26668] *** End of error message ***
Segmentation fault (core dumped)


Thanks for your assistance.



On Sun, Dec 23, 2012 at 4:08 PM, Ralph Castain <rhc@open-mpi.org> wrote:
Do you have your LD_LIBRARY_PATH set correctly? You want your prefix location to be at the front so you don't pickup something from the default system installation.

On Dec 23, 2012, at 12:39 PM, R C <recifx@gmail.com> wrote:

Thanks, r27719 compiles but I get an error when I try:
$/usr/local/bin/ompi_info 
/usr/local/bin/ompi_info: symbol lookup error: /usr/local/bin/ompi_info: undefined symbol: mca_fbtl_base_components_opened

and when I tried to compile a test program:
$ mpicc -o hw hello.c
mpicc: Symbol `opal_show_help' has different size in shared object, consider re-linking



On Sun, Dec 23, 2012 at 2:55 PM, Ralph Castain <rhc@open-mpi.org> wrote:
Sigh - sorry about that. Guess we'll have to feel our way down the various tools. I've committed a fix that -should- resolve this for the rest of the tools in the OMPI layer. Please let me know how it works for you.

Ralph

On Dec 23, 2012, at 10:56 AM, R C <recifx@gmail.com> wrote:


Thanks again for looking into this.
With r27717, 'make all' gives me the error:

Making all in tools/ompi_info
make[2]: Entering directory `/home/rc/Downloads/ompi/objd/ompi/tools/ompi_info'
  CC     ompi_info.o
  CC     param.o
../../../../ompi/tools/ompi_info/param.c: In function ‘ompi_info_do_config’:
../../../../ompi/tools/ompi_info/param.c:238:17: warning: ignoring return value of ‘asprintf’, declared with attribute warn_unused_result [-Wunused-result]
../../../../ompi/tools/ompi_info/param.c:252:17: warning: ignoring return value of ‘asprintf’, declared with attribute warn_unused_result [-Wunused-result]
../../../../ompi/tools/ompi_info/param.c:259:17: warning: ignoring return value of ‘asprintf’, declared with attribute warn_unused_result [-Wunused-result]
../../../../ompi/tools/ompi_info/param.c:271:13: warning: ignoring return value of ‘asprintf’, declared with attribute warn_unused_result [-Wunused-result]
../../../../ompi/tools/ompi_info/param.c:274:13: warning: ignoring return value of ‘asprintf’, declared with attribute warn_unused_result [-Wunused-result]
  CC     components.o
../../../../ompi/tools/ompi_info/components.c: In function ‘ompi_info_register_components’:
../../../../ompi/tools/ompi_info/components.c:106:17: warning: ignoring return value of ‘asprintf’, declared with attribute warn_unused_result [-Wunused-result]
../../../../ompi/tools/ompi_info/components.c:109:21: warning: ignoring return value of ‘asprintf’, declared with attribute warn_unused_result [-Wunused-result]
../../../../ompi/tools/ompi_info/components.c:113:21: warning: ignoring return value of ‘asprintf’, declared with attribute warn_unused_result [-Wunused-result]
  CC     version.o
../../../../ompi/tools/ompi_info/version.c: In function ‘ompi_info_show_ompi_version’:
../../../../ompi/tools/ompi_info/version.c:125:13: warning: ignoring return value of ‘asprintf’, declared with attribute warn_unused_result [-Wunused-result]
../../../../ompi/tools/ompi_info/version.c:134:13: warning: ignoring return value of ‘asprintf’, declared with attribute warn_unused_result [-Wunused-result]
../../../../ompi/tools/ompi_info/version.c:137:13: warning: ignoring return value of ‘asprintf’, declared with attribute warn_unused_result [-Wunused-result]
  CCLD   ompi_info
/usr/bin/ld: version.o: undefined reference to symbol 'orte_info_type_orte'
/usr/bin/ld: note: 'orte_info_type_orte' is defined in DSO /home/rc/Downloads/ompi/objd/orte/.libs/libopen-rte.so.0 so try adding it to the linker command line
/home/rc/Downloads/ompi/objd/orte/.libs/libopen-rte.so.0: could not read symbols: Invalid operation
collect2: error: ld returned 1 exit status
make[2]: *** [ompi_info] Error 1
make[2]: Leaving directory `/home/rc/Downloads/ompi/objd/ompi/tools/ompi_info'
make[1]: *** [all-recursive] Error 1
make[1]: Leaving directory `/home/rc/Downloads/ompi/objd/ompi'
make: *** [all-recursive] Error 1


_______________________________________________
devel mailing list
devel@open-mpi.org
http://www.open-mpi.org/mailman/listinfo.cgi/devel