Ray : Segmentation fault (11)

[Nagesh]

Hello,

I am trying to run Ray to do hybrid assembly of Illumina (paired end data) and Roche 454 data. I am running the program on server with 64 cores and 512GB RAM

I ran the sample data in Ray; which is working fine.
When I give my data, the program is producing Contigs.fasta output but, giving error before scaffolding.

I ran the following command
$ mpiexec -n 20 ray-build/Ray -k 31 -p mp_R1_trimmed_q20_p100_clipped.fastq mp_R2_trimmed_q20_p100_clipped.fastq -s 454.fastq -o kmer_31_out

After contig buildng program gave the following fault while scaffolding

*** Process received signal ***
Signal: Segmentation fault (11)
Signal code: Address not mapped (1)
Failing at address: 0x15b710
[ 0] /lib64/libpthread.so.0() [0x3dbd00f500]
[ 1] ray-build/Ray(_ZNK4Read6lengthEv+0) [0x50a9b0]
[ 2] ray-build/Ray(_ZN16MessageProcessor33call_RAY_MPI_TAG_GET_READ_MARKERSEP7Message+0x9f) [0x49c09f]
[ 3] ray-build/Ray(_ZN11ComputeCore10runVanillaEv+0x61) [0x532351]
[ 4] ray-build/Ray(_ZN11ComputeCore3runEv+0x6e) [0x53379e]
[ 5] ray-build/Ray(_ZN7Machine5startEv+0x1697) [0x4487a7]
[ 6] ray-build/Ray(main+0x3a) [0x443bba]
[ 7] /lib64/libc.so.6(__libc_start_main+0xfd) [0x3dbc41ecdd]
[ 8] ray-build/Ray() [0x443a89]
*** End of error message ***
--------------------------------------------------------------------------
mpiexec noticed that process rank 19 with PID 21982 on node pragnya exited on signal 11 (Segmentation fault).
--------------------------------------------------------------------------

I tried to run the program with different number of ranks and kmers. But still stoping at scaffolding stage.

Can anyone please help

[raw937]

Ray:segmentation fault (11)

So, a question as
I am having the same problem.
I think its an open mpi error.
Run Top - do you see if all 20 cores are running Ray?

In my case, only one is running causing the error.

Double check and get back to me?

cheers
Rick

[Nagesh]

RE : Ray : Segmentation fault (11)

Hi,

As you suggested, it is an open mpi fault. But when I run the program I see 20 cores being used for initial period of about 1 hour. The program stops after the contig building. The contigs are not scaffolded because of this error

[dorix]

Hi,
I have the same problem. Has anyone managed to solve it?

Interesting fact is that I have managed to finish many samples with the same machine, mpi, settings, comparable complexity and size... This one refuses to cooperate.

Regards