Bugzilla – Bug 3589
pre-WS GT4 globusrun -a mode cores on a USATLAS site
Last modified: 2005-07-22 14:51:36
You need to log in before you can comment on or make changes to this bug.
I have one peculiar site on OSG that I cannot test authentication with using the pre-WS globusrun shipped in GT4.0.1: $ globusrun -a -r tier2.uchicago.edu/jobmanager-fork Segmentation fault (core dumped) $ type globusrun globusrun is hashed (/opt/globus/latest/bin/globusrun) Using my old vanilla GT2, it works well with the site: $ LD_LIBRARY_PATH=/vdt/globus/lib /vdt/globus/bin/globusrun -a -r tier2.uchicago.edu/jobmanager-fork GRAM Authentication test successful Here is the trace from the core file: #0 0x40403520 in _int_free () from /lib/i686/libc.so.6 #1 0x4040239c in free () from /lib/i686/libc.so.6 #2 0x40132a11 in gss_compare_name (minor_status=0xbfffce10, name1_P=0x806b840, name2_P=0x806bd38, name_equal=0xbfffce04) at compare_name.c:211 #3 0x401361ab in gss_init_sec_context (minor_status=0xbfffceb4, initiator_cred_handle=0x805ec18, context_handle_P=0x806dce8, target_name=0x806bd38, mech_type=0x0, req_flags=34, time_req=0, input_chan_bindings=0x0, input_token=0xbfffcea0, actual_mech_type=0x806dcf4, output_token=0xbfffcea8, ret_flags=0x806dcdc, time_rec=0x806dce0) at init_sec_context.c:242 #4 0x400d2729 in globus_l_xio_gsi_read_token_cb (op=0x806db70, result=0, nbytes=71, user_arg=0x806dcd8) at globus_xio_gsi.c:1136 #5 0x400b1427 in globus_l_xio_driver_op_read_kickout (user_arg=0x806db70) at globus_xio_driver.c:620 #6 0x400c1618 in globus_xio_driver_finished_read (in_op=0x806db70, result=0, nbytes=71) at globus_xio_pass.c:1227 #7 0x400f389a in globus_l_xio_tcp_finish_read (handle=0x806b8b8, result=0, nbytes=71) at globus_xio_tcp_driver.c:2196 #8 0x400f3923 in globus_l_xio_tcp_system_read_cb (result=0, nbytes=71, user_arg=0x806b8b8) at globus_xio_tcp_driver.c:2211 #9 0x400c4c5d in globus_l_xio_system_kickout (user_arg=0x8086f60) at globus_xio_system_select.c:1016 #10 0x401c2d80 in globus_callback_space_poll (timestop=0x8050750, space=-2) at globus_callback_nothreads.c:1424 #11 0x40030e03 in globus_gram_client_ping ( resource_manager_contact=0x806be08 "tier2.uchicago.edu/jobmanager-fork") at globus_gram_client.c:747 #12 0x0804af19 in main (argc=1, argv=0xbfffd214) at globusrun.c:660 #13 0x403a4a67 in __libc_start_main () from /lib/i686/libc.so.6 The Globus version is a CVS checkout 4.0.0 as of 20050616. The system is a RH9 Linux.
Jens, We are in the process of testing release candidates for 4.0.1. If you could rebuild with rc1 on that host and let us know if this is still a problem, then Joe can take a look at it. -Stu
This is probably a dupe of http://bugzilla.globus.org/globus/show_bug.cgi? id=3564 which has been fixed.
Yes, that could very well be. Using the site's real name tier2-01 instead of tier2, it does not die on a core. The compilation will see, if it is fixed in rc1.
OK, with 4.0.1-rc1 the core dumps are gone.
I won't mark this as a dup of 3564, since we are not sure. I'll just mark it fixed. Thanks Jens!