Application Ended Properly with Error "Failed to Bind Memory"
Symptom
Running an MPI job on one or more nodes with the specified -bind-to core triggers the error "failed to bind memory".
-------------------------------------------------------------------------- WARNING: Open MPI tried to bind a process but failed. This is a warning only; your job will continue, though performance may be degraded. Local host: node175 Application name: /root/tqa/osu-micro-benchmarks-7.1-1/c/mpi/collective/blocking/osu_bcast Error message: failed to bind memory Location: rtc_hwloc.c:447 -------------------------------------------------------------------------- [node175:06049] 63 more processes have sent help message help-orte-odls-default.txt / memory not bound [node175:06049] Set MCA parameter "orte_base_help_aggregate" to 0 to see all help / error messages
Possible Cause
The node is not fully inserted with memory modules, leaving some NUMA nodes unbound. To verify if memory modules are bound, run the numactl --hardware command.
available: 4 nodes (0-3) node 0 cpus: 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 node 0 size: 31566 MB node 0 free: 30548 MB node 1 cpus: 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 node 1 size: 0 MB node 1 free: 0 MB node 2 cpus: 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 node 2 size: 31049 MB node 2 free: 29424 MB node 3 cpus: 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 node 3 size: 0 MB node 3 free: 0 MB node distances: node 0 1 2 3 0: 10 12 35 37 1: 12 10 37 40 2: 35 37 10 12 3: 37 40 12 10
The above information shows that neither node1 nor node2 is bound to the memory.
Recovery Procedure
- For smaller-scale processes, use rankfile to avoid binding processes to NUMA nodes lacking memory.
- Insert DIMMs fully.
Parent topic: FAQ