The slots are numbered 0-3 for a four-slot allocation as shown in your PE_HOSTFILE. Your rankfile contains assignments to slot=4 and slot=5, which are outside your allocation.
On Apr 23, 2010, at 10:36 AM, Orion Poplawski wrote: > I'm using gridengine 6.2u5 and openmpi 1.3.3. I'm submitting a parallel > job and would like to specify a rankfile to set processor binding but am > getting errors. > > The $PE_HOSTFILE generated by gridengine is: > > amos.cora.nwra.com 4 cloud...@amos.cora.nwra.com UNDEFINED > andrew.cora.nwra.com 4 cloud...@andrew.cora.nwra.com UNDEFINED > > The rankfile I'm using is: > > rank 0=amos.cora.nwra.com slot=0 > rank 1=andrew.cora.nwra.com slot=0 > rank 2=amos.cora.nwra.com slot=4 > rank 3=andrew.cora.nwra.com slot=4 > rank 4=amos.cora.nwra.com slot=1 > rank 5=andrew.cora.nwra.com slot=1 > rank 6=amos.cora.nwra.com slot=5 > rank 7=andrew.cora.nwra.com slot=5 > > The error I'm getting is: > > Rankfile claimed host amos.cora.nwra.com that was not allocated or > oversubscribed it's slots: > > -------------------------------------------------------------------------- > [amos:05727] [[44126,0],0] ORTE_ERROR_LOG: Bad parameter in file > rmaps_rank_file.c at line 108 > [amos:05727] [[44126,0],0] ORTE_ERROR_LOG: Bad parameter in file > base/rmaps_base_map_job.c at line 87 > [amos:05727] [[44126,0],0] ORTE_ERROR_LOG: Bad parameter in file > base/plm_base_launch_support.c at line 77 > [amos:05727] [[44126,0],0] ORTE_ERROR_LOG: Bad parameter in file > plm_rsh_module.c at line 990 > -------------------------------------------------------------------------- > A daemon (pid unknown) died unexpectedly on signal 1 while attempting to > launch so we are aborting. > > Any ideas? > > Thanks! > > - Orion > > -- > Orion Poplawski > Technical Manager 303-415-9701 x222 > NWRA/CoRA Division FAX: 303-415-9702 > 3380 Mitchell Lane or...@cora.nwra.com > Boulder, CO 80301 http://www.cora.nwra.com > _______________________________________________ > users mailing list > us...@open-mpi.org > http://www.open-mpi.org/mailman/listinfo.cgi/users
