Hello, i've found that on my system mpirun /bin/hostname causes segmentation fault. Inspecting it in gdb, i've seen the following stack trace: (gdb) run /bin/hostname Starting program: /usr/bin/mpirun /bin/hostname [Thread debugging using libthread_db enabled] Using host libthread_db library "/lib64/libthread_db.so.1". Program received signal SIGSEGV, Segmentation fault. 0x00007ffff69c484a in ?? () from /lib64/libc.so.6 (gdb) bt #0 0x00007ffff69c484a in ?? () from /lib64/libc.so.6 #1 0x0000000000430455 in HYDTI_bscd_in_env_list (env_name=0x0, env_list=env_list@entry=0x7fffffffd890) at tools/bootstrap/external/external_common.c:27 #2 0x0000000000425ae4 in HYDT_bscd_ssh_query_env_inherit (env_name=<optimized out>, ret=0x7fffffffd8e4) at tools/bootstrap/external/ssh_env.c:17 #3 0x000000000042557c in HYDT_bsci_query_env_inherit (env_name=<optimized out>, ret=ret@entry=0x7fffffffd8e4) at tools/bootstrap/src/bsci_env.c:18 #4 0x00000000004208d3 in HYDU_list_inherited_env (env_list=env_list@entry=0x66c680 <HYD_server_info+128>) at utils/env/env.c:55 #5 0x000000000040404c in main (argc=<optimized out>, argv=<optimized out>) at ui/mpich/mpiexec.c:258 The problem seems to be in strcmp call, which compares NULL (empty named environment variable) with something. env command shows string "=" for me. So, I guess it would be nice to check env_name variable before passing it to strcmp? -- Regards, Dmitriy