AMBER Archive (2003)

Subject: mpi/sander help

From: Thu Zar Lwin (tlwin_at_ea.nacs.uci.edu)
Date: Fri Feb 07 2003 - 16:01:05 CST


HI,
  I run a mpi job using sander on 16 nodes (one node being the Master and
the other 15 nodes being the slaves) I got the following error message. I
am wondering what I should do to eliminate those errors.

p13_15738: p4_error: net_recv read: probable EOF on socket: 1
p6_961: p4_error: interrupt SIGSEGV: 11
p7_322: p4_error: net_recv read: probable EOF on socket: 1
bm_list_28984: (5562.725016) wakeup_slave: unable to interrupt slave 0 pid
28983
p11_32275: p4_error: net_recv read: probable EOF on socket: 1
bm_list_28984: (5562.810537) wakeup_slave: unable to interrupt slave 0 pid
28983
rm_l_1_29553: (5562.461563) net_send: could not write to fd=6, errno = 9
rm_l_1_29553: p4_error: net_send write: -1
p10_4792: p4_error: net_recv read: probable EOF on socket: 1
bm_list_28984: (5562.817421) wakeup_slave: unable to interrupt slave 0 pid
28983
p9_6687: p4_error: net_recv read: probable EOF on socket: 1
bm_list_28984: (5562.831913) wakeup_slave: unable to interrupt slave 0 pid
28983
p14_1810: p4_error: net_recv read: probable EOF on socket: 1
bm_list_28984: (5562.842485) wakeup_slave: unable to interrupt slave 0 pid
28983
p3_8794: p4_error: interrupt SIGSEGV: 11
p5_31353: p4_error: interrupt SIGSEGV: 11
p2_28479: p4_error: net_recv read: probable EOF on socket: 1
bm_list_28984: (5562.941301) wakeup_slave: unable to interrupt slave 0 pid
28983
p15_7002: p4_error: net_recv read: probable EOF on socket: 1
bm_list_28984: (5562.964986) wakeup_slave: unable to interrupt slave 0 pid
28983
p4_11689: p4_error: interrupt SIGSEGV: 11
p8_12757: p4_error: net_recv read: probable EOF on socket: 1
bm_list_28984: (5562.983074) wakeup_slave: unable to interrupt slave 0 pid
28983
p12_19241: p4_error: net_recv read: probable EOF on socket: 1
bm_list_28984: (5563.037984) wakeup_slave: unable to interrupt slave 0 pid
28983
Broken pipe
Broken pipe
Broken pipe
Broken pipe
Broken pipe
Broken pipe

Thanks,
Thu