j-woz / exm-issues

Automatically exported from code.google.com/p/exm-issues
0 stars 0 forks source link

PMPI_Comm_create_group error with parallel tasks #577

Open GoogleCodeExporter opened 9 years ago

GoogleCodeExporter commented 9 years ago
When doing some testing of very short parallel tasks, I ran into this problem.

I created a Turbine test ptasks-stress to replicate it.

[1] Fatal error in PMPI_Comm_create_group: Other MPI error, error stack:
[1] PMPI_Comm_create_group(230)..........: 
MPI_Comm_create_group(comm=0x84000002, group=0xc80013bb, 0, 
new_comm=0x7fff1e428a00) failed
[1] PMPI_Comm_create_group(212)..........:
[1] MPIR_Comm_create_group(64)...........:
[1] MPIR_Get_contextid_sparse_group(1146):  Cannot allocate context ID because 
of fragmentation (57/2048 free on this process; ignore_id=0)
[2] Fatal error in PMPI_Comm_create_group: Other MPI error, error stack:
[2] PMPI_Comm_create_group(230)..........: 
MPI_Comm_create_group(comm=0x84000002, group=0xc80013da, 0, 
new_comm=0x7fff705ac2e0) failed
[2] PMPI_Comm_create_group(212)..........:
[2] MPIR_Comm_create_group(64)...........:
[2] MPIR_Get_contextid_sparse_group(1146):  Cannot allocate context ID because 
of fragmentation (26/2048 free on this process; ignore_id=0)

Original issue reported on code.google.com by tim.g.ar...@gmail.com on 21 Nov 2013 at 2:39

GoogleCodeExporter commented 9 years ago
Should we post this to the MPICH list?

Original comment by wozniak....@gmail.com on 13 Dec 2013 at 4:54

GoogleCodeExporter commented 9 years ago
I have absolutely no idea what the problem is, so maybe :)

I'm not sure if this is an internal MPICH bug or us somehow misusing the API.

Original comment by tim.g.ar...@gmail.com on 13 Dec 2013 at 4:57

GoogleCodeExporter commented 9 years ago

Original comment by tim.g.ar...@gmail.com on 13 Jun 2014 at 1:56

GoogleCodeExporter commented 9 years ago

Original comment by wozniak....@gmail.com on 6 Nov 2014 at 4:31

GoogleCodeExporter commented 9 years ago
I will be looking at complex parallel tasks again soon.

Original comment by wozniak....@gmail.com on 12 Mar 2015 at 2:30